Prometheus Metrics for On-prem Users
BuildBuddy exposes Prometheus metrics that allow monitoring the four golden signals: latency, traffic, errors, and saturation.
Prometheus metrics are exposed under the path metrics/ on port 9090 by default.
To view these metrics in a live-updating dashboard, we recommend using a tool like Grafana.
Invocation build event metrics
All invocation metrics are recorded at the end of each invocation.
buildbuddy_invocation_count (Counter)
The total number of invocations whose logs were uploaded to BuildBuddy.
Labels
- invocation_status: Invocation status:
success,failure,disconnected, orunknown. - bazel_exit_code: Exit code of a completed bazel command
- bazel_command: Command provided to the Bazel daemon:
run,test,build,coverage,mobile-install, ...
Examples
# Number of invocations per second by invocation status
sum by (invocation_status) (rate(buildbuddy_invocation_count[5m]))
# Invocation success rate
sum(rate(buildbuddy_invocation_count{invocation_status="success"}[5m]))
/
sum(rate(buildbuddy_invocation_count[5m]))
buildbuddy_invocation_log_live_chunk_written_bytes (Counter)
Total number of bytes written to the key-value store for live (in-progress) invocation log tail chunks.
Labels
- suffix_writes_enabled: Whether live invocation log chunks were written to the key-value store with suffix-only writes:
trueorfalse(experiment arm).
buildbuddy_invocation_duration_usec (Histogram)
The total duration of each invocation, in microseconds.
Labels
- invocation_status: Invocation status:
success,failure,disconnected, orunknown. - bazel_command: Command provided to the Bazel daemon:
run,test,build,coverage,mobile-install, ...
Examples
# Median invocation duration in the past 5 minutes
histogram_quantile(
0.5,
sum(rate(buildbuddy_invocation_duration_usec_bucket[5m])) by (le)
)
buildbuddy_invocation_open_streams (Gauge)
Number of build event streams currently being handled by the server.
buildbuddy_invocation_build_event_count (Counter)
Number of build events uploaded to BuildBuddy.
Labels
- status: Status code as defined by grpc/codes. This is a numeric value; any non-zero code indicates an error.
Examples
# Build events uploaded per second
sum(rate(buildbuddy_invocation_build_event_count[5m]))
# Approximate error rate of build event upload handler
sum(rate(buildbuddy_invocation_build_event_count{status="0"}[5m]))
/
sum(rate(buildbuddy_invocation_build_event_count[5m]))
buildbuddy_invocation_stats_recorder_workers (Gauge)
Number of invocation stats recorder workers currently running.
buildbuddy_invocation_stats_recorder_duration_usec (Histogram)
How long it took to finalize an invocation's stats, in microseconds.
This includes the time required to wait for all BuildBuddy apps to flush their local metrics to Redis (if applicable) and then record the metrics to the DB.
buildbuddy_invocation_webhook_invocation_lookup_workers (Gauge)
Number of webhook invocation lookup workers currently running.
buildbuddy_invocation_webhook_invocation_lookup_duration_usec (Histogram)
How long it took to lookup an invocation before posting to the webhook, in microseconds.
buildbuddy_invocation_webhook_invocation_payload_size_bytes (Histogram)
Size in bytes of the payload posted to invocation webhook endpoints.
buildbuddy_invocation_webhook_notify_workers (Gauge)
Number of webhook notify workers currently running.
buildbuddy_invocation_webhook_notify_duration_usec (Histogram)
How long it took to post an invocation proto to the webhook, in microseconds.
Remote cache metrics
NOTE: Cache metrics are recorded at the end of each invocation, which means that these metrics provide approximate real-time signals.
buildbuddy_remote_cache_events (Counter)
Number of cache events handled.
Labels
- cache_type: Cache type:
actionfor action cache,casfor content-addressable storage. - cache_event_type: Cache event type:
hit,miss, orupload. - group_id: Group (organization) ID associated with the request.
- tracked: Whether or not billable usage was recorded for this request ("true", "false")
buildbuddy_remote_cache_download_size_bytes (Histogram)
Number of bytes downloaded from the remote cache in each download.
Use the _sum suffix to get the total downloaded bytes and the _count suffix to get the number of downloaded files.
Labels
- cache_type: Cache type:
actionfor action cache,casfor content-addressable storage. - server_name: Describes the name of the server that handles a client request, such as "byte_stream_server" or "cas_server"
- group_id: Group (organization) ID associated with the request.
- tracked: Whether or not billable usage was recorded for this request ("true", "false")
Examples
# Cache download rate (bytes per second)
sum(rate(buildbuddy_cache_download_size_bytes_sum[5m]))
buildbuddy_remote_cache_download_duration_usec (Histogram)
Download duration for each file downloaded from the remote cache, in microseconds.
Labels
- cache_type: Cache type:
actionfor action cache,casfor content-addressable storage. - tracked: Whether or not billable usage was recorded for this request ("true", "false")
Examples
# Median download duration for content-addressable store (CAS)
histogram_quantile(
0.5,
sum(rate(buildbuddy_remote_cache_download_duration_usec{cache_type="cas"}[5m])) by (le)
)
buildbuddy_remote_cache_upload_size_bytes (Histogram)
Number of bytes uploaded to the remote cache in each upload.
Use the _sum suffix to get the total uploaded bytes and the _count suffix to get the number of uploaded files.
Labels
- cache_type: Cache type:
actionfor action cache,casfor content-addressable storage. - server_name: Describes the name of the server that handles a client request, such as "byte_stream_server" or "cas_server"
- group_id: Group (organization) ID associated with the request.
- tracked: Whether or not billable usage was recorded for this request ("true", "false")
Examples
# Cache upload rate (bytes per second)
sum(rate(buildbuddy_cache_upload_size_bytes_sum[5m]))
buildbuddy_remote_cache_upload_duration_usec (Histogram)
Upload duration for each file uploaded to the remote cache, in microseconds.
Labels
- cache_type: Cache type:
actionfor action cache,casfor content-addressable storage. - tracked: Whether or not billable usage was recorded for this request ("true", "false")
Examples
# Median upload duration for content-addressable store (CAS)
histogram_quantile(
0.5,
sum(rate(buildbuddy_remote_cache_upload_duration_usec{cache_type="cas"}[5m])) by (le)
)
buildbuddy_remote_cache_method_handled_total (Counter)
Total number of methods completed on the cache, regardless of success or failure.
Labels
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
- cache_method: The Methods exposed in interfaces.Cache.
- status: Status code as defined by grpc/codes in human-readable format, such as "OK" or "NotFound".
buildbuddy_remote_cache_method_handling_usec (Histogram)
Histogram of response latency(microseconds) of cache methods
Labels
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
- cache_method: The Methods exposed in interfaces.Cache.
buildbuddy_remote_cache_disk_cache_last_eviction_age_usec (Gauge)
The age of the item most recently evicted from the cache, in microseconds.
Labels
- partition_id: The ID of the disk cache partition this event applied to.
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
buildbuddy_remote_cache_disk_cache_eviction_age_msec (Histogram)
Age of items evicted from the cache, in milliseconds.
Labels
- partition_id: The ID of the disk cache partition this event applied to.
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
buildbuddy_remote_cache_disk_cache_num_evictions (Counter)
Number of items evicted.
Labels
- partition_id: The ID of the disk cache partition this event applied to.
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
buildbuddy_remote_cache_disk_cache_partition_size_bytes_evicted (Counter)
Number of bytes in the partition evicted.
Labels
- partition_id: The ID of the disk cache partition this event applied to.
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
buildbuddy_remote_cache_disk_cache_partition_size_bytes (Gauge)
Number of bytes in the partition.
Labels
- partition_id: The ID of the disk cache partition this event applied to.
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
buildbuddy_remote_cache_disk_cache_sampled_partition_group_size_bytes (Gauge)
Number of bytes seen while sampling the partition for eviction, by group ID.
Labels
- partition_id: The ID of the disk cache partition this event applied to.
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
- group_id: Group (organization) ID associated with the request.
buildbuddy_remote_cache_disk_cache_partition_capacity_bytes (Gauge)
Number of bytes in the partition.
Labels
- partition_id: The ID of the disk cache partition this event applied to.
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
buildbuddy_remote_cache_disk_cache_partition_num_items (Gauge)
Number of items in the partition.
Labels
- partition_id: The ID of the disk cache partition this event applied to.
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
- cache_type: Cache type:
actionfor action cache,casfor content-addressable storage.
buildbuddy_remote_cache_disk_cache_duplicate_writes (Counter)
Number of writes for digests that already exist.
Labels
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
buildbuddy_remote_cache_disk_cache_added_file_size_bytes (Histogram)
Size of artifacts added to the file cache, in bytes.
Labels
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
buildbuddy_remote_cache_disk_cache_filesystem_total_bytes (Gauge)
Total size of the underlying filesystem.
Labels
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
buildbuddy_remote_cache_disk_cache_filesystem_avail_bytes (Gauge)
Available bytes in the underlying filesystem.
Labels
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
Examples
# Total number of duplicate writes.
sum(buildbuddy_remote_cache_duplicate_writes)
buildbuddy_remote_cache_disk_cache_duplicate_writes_bytes (Counter)
Number of bytes written that already existed in the cache.
Labels
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
buildbuddy_remote_cache_distributed_cache_peer_lookups (Histogram)
Number of peers consulted (including the 'local peer') for a distributed cache read before returning a response.
For batch requests, one observation is recorded for each digest in the request.
Labels
- op: Distributed cache operation name, such as "FindMissing" or "Get".
- cache_status: Cache lookup result - One of: - "hit" - "miss" - "partial" (for batched RPCs where part of a request was cached) - Or "uncacheable" (for e.g. encrypted resources)
buildbuddy_remote_cache_distributed_cache_find_missing_blob_status_count (Counter)
Count of blobs checked by FindMissing at the distributed-cache layer, by logical present/absent status and originating purpose.
Labels
- purpose: FindMissing purpose: which internal code path originated a FindMissing lookup (see repb.FindMissingBlobsRequest.Purpose), e.g. "ATIME_UPDATE".
- status: Status code as defined by grpc/codes. This is a numeric value; any non-zero code indicates an error.
buildbuddy_remote_cache_distributed_cache_backfill_count (Counter)
Count of distributed cache backfills, by whether the blob was sent to the peer as a reference or as inline bytes, and by status code.
Labels
- request_type: How a distributed cache write's payload was sent: "reference" (a pointer to the blob in shared storage) or "bytes" (the blob's bytes, streamed inline).
- status: Status code as defined by grpc/codes in human-readable format, such as "OK" or "NotFound".
buildbuddy_remote_cache_distributed_cache_backfill_size_bytes (Counter)
Total digest sizes of blobs backfilled to distributed cache peers, by whether the blob was sent as a reference or as inline bytes, and by status code.
Labels
- request_type: How a distributed cache write's payload was sent: "reference" (a pointer to the blob in shared storage) or "bytes" (the blob's bytes, streamed inline).
- status: Status code as defined by grpc/codes in human-readable format, such as "OK" or "NotFound".
buildbuddy_remote_cache_distributed_cache_backfill_latency_usec (Histogram)
Latency of backfilling a single digest.
Labels
- group_id: Group (organization) ID associated with the request.
- status: Status code as defined by grpc/codes. This is a numeric value; any non-zero code indicates an error.
buildbuddy_remote_cache_distributed_cache_read_response_count (Counter)
Count of distributed cache peer reads, by whether the payload was received as a reference or as inline bytes, and by status code.
Labels
- response_type: How a distributed cache read's payload was received: "reference" (a pointer to the blob in shared storage) or "bytes" (the blob's bytes, streamed inline).
- status: Status code as defined by grpc/codes in human-readable format, such as "OK" or "NotFound".
buildbuddy_remote_cache_distributed_cache_read_response_size_bytes (Counter)
Total digest sizes of blobs read from distributed cache peers, by whether the payload was received as a reference or as inline bytes, and by status code.
Labels
- response_type: How a distributed cache read's payload was received: "reference" (a pointer to the blob in shared storage) or "bytes" (the blob's bytes, streamed inline).
- status: Status code as defined by grpc/codes in human-readable format, such as "OK" or "NotFound".
buildbuddy_remote_cache_distributed_cache_write_request_count (Counter)
Count of distributed cache peer writes, by whether the payload was sent as a reference or as inline bytes, and by status code.
Labels
- request_type: How a distributed cache write's payload was sent: "reference" (a pointer to the blob in shared storage) or "bytes" (the blob's bytes, streamed inline).
- status: Status code as defined by grpc/codes in human-readable format, such as "OK" or "NotFound".
buildbuddy_remote_cache_distributed_cache_write_request_size_bytes (Counter)
Total digest sizes of blobs written to distributed cache peers, by whether the payload was sent as a reference or as inline bytes, and by status code.
Labels
- request_type: How a distributed cache write's payload was sent: "reference" (a pointer to the blob in shared storage) or "bytes" (the blob's bytes, streamed inline).
- status: Status code as defined by grpc/codes in human-readable format, such as "OK" or "NotFound".
buildbuddy_remote_cache_distributed_cache_reference_verification_count (Counter)
Count of reference verifications on distributed cache reads, by group, outcome, and error code.
Labels
- group_id: Group (organization) ID associated with the request.
- outcome: The outcome of a verification operation. One of: ["success", "failure", "error"]
- status: Status code as defined by grpc/codes in human-readable format, such as "OK" or "NotFound".
buildbuddy_remote_cache_distributed_cache_reference_write_verification_count (Counter)
Count of reference verifications on distributed cache writes, by group, outcome, and error code.
Labels
- group_id: Group (organization) ID associated with the request.
- outcome: The outcome of a verification operation. One of: ["success", "failure", "error"]
- status: Status code as defined by grpc/codes in human-readable format, such as "OK" or "NotFound".
buildbuddy_remote_cache_migration_not_found_error_count (Counter)
Number of not found errors from the destination cache during a cache migration.
Labels
- type: Describes the type of cache request
- group_id: Group (organization) ID associated with the request.
buildbuddy_remote_cache_migration_double_read_hit_count (Counter)
Number of double reads where the source and destination caches hold the same digests during a cache migration.
Labels
- type: Describes the type of cache request
- group_id: Group (organization) ID associated with the request.
buildbuddy_remote_cache_migration_copy_chan_size (Gauge)
Number of digests queued to be copied during a cache migration.
buildbuddy_remote_cache_migration_bytes_copied (Counter)
Number of bytes copied from the source to destination cache during a cache migration.
Labels
- cache_type: Cache type:
actionfor action cache,casfor content-addressable storage. - group_id: Group (organization) ID associated with the request.
buildbuddy_remote_cache_migration_blobs_copied (Counter)
Number of blobs copied from the source to destination cache during a cache migration.
Labels
- cache_type: Cache type:
actionfor action cache,casfor content-addressable storage. - group_id: Group (organization) ID associated with the request.
buildbuddy_remote_cache_tree_cache_lookup_count (Counter)
Total number of TreeCache lookups.
Labels
- status: The TreeCache status: hit/miss/invalid_entry.
- level: TreeCache directory depth: 0 for the root dir, 1 for a direct child of the root dir, and so on.
buildbuddy_remote_cache_tree_cache_split_lookup_count (Counter)
Total number of TreeCache split lookups.
Labels
- status: The TreeCache split lookup status: hit/miss/failure
buildbuddy_remote_cache_tree_cache_split_write_count (Counter)
Total number of splits written to TreeCache.
buildbuddy_remote_cache_tree_cache_set_count (Counter)
Total number of TreeCache sets.
Labels
- status: The TreeCache set status: success/deadline_exceeded/other_error
buildbuddy_remote_cache_tree_cache_bytes_transferred (Counter)
Number of bytes written or read from tree cache
Labels
- op: TreeCache operation "read" or "write"
buildbuddy_remote_cache_splice_blob_duration_usec (Histogram)
Duration of the full SpliceBlob RPC handler, in microseconds.
Labels
- status: Status code as defined by grpc/codes. This is a numeric value; any non-zero code indicates an error.
buildbuddy_remote_cache_splice_blob_count (Counter)
Number of SpliceBlob requests that reached the manifest store step, labeled by whether chunk contents were verified against the blob digest.
Labels
- splice_validation: Whether a SpliceBlob request hashed the referenced chunks to verify the blob digest (
full) or skipped that check (skipped). - group_id: Group (organization) ID associated with the request.
buildbuddy_remote_cache_chunked_manifest_validation_count (Counter)
Number of chunked manifest Store() calls, labeled by whether the shared validation marker was a hit (skipped re-hashing) or miss (full chunk verification performed).
Labels
- cache_status: Cache lookup result - One of: - "hit" - "miss" - "partial" (for batched RPCs where part of a request was cached) - Or "uncacheable" (for e.g. encrypted resources)
buildbuddy_remote_cache_chunked_manifest_load_count (Counter)
Number of successful chunked manifest LoadManifest() calls, labeled by manifest key scheme.
Labels
- prefix:
buildbuddy_remote_execution_get_tree_directory_lookup_count (Counter)
Number of directories fetched by GetTree calls, split by where the directory was found.
Labels
- location: Where a directory from a GetTree request was found: one of "uncached", "filecache", "remote"
buildbuddy_remote_execution_get_tree_filecache_trees_read (Counter)
Number of trees read from the local filecache.
buildbuddy_remote_execution_get_tree_filecache_bytes_read (Counter)
Total size in bytes of trees read from the local filecache.
buildbuddy_remote_execution_get_tree_filecache_trees_written (Counter)
Number of trees written to the local filecache.
buildbuddy_remote_execution_get_tree_filecache_bytes_written (Counter)
Total size in bytes of trees written to the local filecache.
buildbuddy_remote_cache_lookaside_cache_lookup_count (Counter)
Total number of Lookaside Cache lookups.
Labels
- status: The Lookaside cache status: hit/miss.
buildbuddy_remote_cache_lookaside_cache_lookup_bytes (Counter)
Total number bytes served from the Lookaside Cache by hit/miss status.
Labels
- status: The Lookaside cache status: hit/miss.
buildbuddy_remote_cache_lookaside_cache_eviction_age_msec (Histogram)
Age of items evicted from the cache, in milliseconds.
Labels
- eviction_reason: The reason an item was evicted from the lookaside cache. One of: "expired" or "size"
Remote execution metrics
buildbuddy_remote_execution_count (Counter)
Number of actions executed remotely.
This only includes actions which reached the execution phase. If an action fails before execution (for example, if it fails authentication) then this metric is not incremented.
Labels
- exit_code: Process exit code of an executed action.
- status: Status code as defined by grpc/codes in human-readable format, such as "OK" or "NotFound".
- isolation: Effective workload isolation type used for an executed task, such as "docker", "podman", "firecracker", or "none".
Examples
# Total number of actions executed per second
sum(rate(buildbuddy_remote_execution_count[5m]))
buildbuddy_remote_execution_tasks_started_count (Counter)
Number of tasks started remotely, but not necessarily completed.
Includes retry attempts of the same task.
buildbuddy_remote_execution_executed_action_metadata_durations_usec (Histogram)
Time spent in each stage of action execution, in microseconds.
Queries should filter or group by the stage label, taking care not to aggregate different stages.
Labels
- stage: Executed action stage. Action execution is split into stages corresponding to the timestamps defined in
ExecutedActionMetadata:queued,input_fetch,execution, andoutput_upload. An additional stage,worker, includes all stages during which a worker is handling the action, which is all stages except thequeuedstage. - group_id: Group (organization) ID associated with the request.
Examples
# Median duration of all command stages
histogram_quantile(
0.5,
sum(rate(buildbuddy_remote_execution_executed_action_metadata_durations_usec_bucket[5m])) by (le, stage)
)
# p90 duration of just the command execution stage
histogram_quantile(
0.9,
sum(rate(buildbuddy_remote_execution_executed_action_metadata_durations_usec_bucket{stage="execution"}[5m])) by (le)
)
buildbuddy_remote_execution_task_pressure_stall_duration_fraction (Histogram)
Linux PSI stall time as a fraction of each action's execution duration (0-1).
Labels
- resource: System resource: "cpu", "memory", or "io".
- stall_type: Pressure stall type: "some" (task is partially stalled on the resource) or "full" (task is completely stalled on the resource).
buildbuddy_remote_execution_task_size_read_requests (Counter)
Number of read requests to the task sizer, which estimates action resource usage based on historical execution stats.
Labels
- status: Status of the task size read request:
hit,miss, orerror. - isolation: Effective workload isolation type used for an executed task, such as "docker", "podman", "firecracker", or "none".
- os: OS associated with the request.
- arch: CPU architecture associated with the request.
- group_id: Group (organization) ID associated with the request.
buildbuddy_remote_execution_task_size_write_requests (Counter)
Number of write requests to the task sizer, which estimates action resource usage based on historical execution stats.
Labels
- status: Status of the task size write request:
ok,missing_statsorerror. - isolation: Effective workload isolation type used for an executed task, such as "docker", "podman", "firecracker", or "none".
- os: OS associated with the request.
- arch: CPU architecture associated with the request.
- group_id: Group (organization) ID associated with the request.
buildbuddy_remote_execution_task_size_prediction_duration_usec (Histogram)
Task size prediction model request duration in microseconds.
Labels
- status: Status code as defined by grpc/codes in human-readable format, such as "OK" or "NotFound".
buildbuddy_remote_execution_buildroot_disk_usage_measurement_duration_usec (Histogram)
Duration of the buildroot (workspace) disk usage measurement performed after a task finishes, in microseconds.
buildbuddy_remote_execution_resource_usage_timeline_metadata_size_bytes (Histogram)
Total size of resource usage timeline payloads received from executors.
Labels
- group_id: Group (organization) ID associated with the request.
buildbuddy_remote_execution_input_download_bitmap_metadata_size_bytes (Histogram)
Total size of input download bitmaps received from executors.
Labels
- group_id: Group (organization) ID associated with the request.
buildbuddy_remote_execution_enqueued_task_milli_cpu (Histogram)
Milli-CPU prediction of enqueued tasks.
buildbuddy_remote_execution_enqueued_task_memory_bytes (Histogram)
Memory prediction of enqueued tasks.
buildbuddy_remote_execution_oom_killer_targeted_task_memory_bytes (Histogram)
Observed task memory in bytes targeted by the executor OOM killer.
Labels
- group_id: Group (organization) ID associated with the request.
buildbuddy_remote_execution_waiting_execution_result (Gauge)
Number of execution requests for which the client is actively waiting for results.
Labels
- group_id: Group (organization) ID associated with the request.
Examples
# Total number of execution requests with client waiting for result.
sum(buildbuddy_remote_execution_waiting_execution_result)
buildbuddy_remote_execution_requests (Counter)
Number of execution requests received.
Labels
- group_id: Group (organization) ID associated with the request.
- os: OS associated with the request.
- arch: CPU architecture associated with the request.
buildbuddy_remote_execution_executor_registration_count (Counter)
Number of executor registrations on the scheduler.
Labels
- version: Binary version. Example:
v2.0.0.
Examples
# Rate of new execution requests by OS/Arch.
sum(rate(buildbuddy_remote_execution_requests[1m])) by (os, arch)
buildbuddy_remote_execution_merged_actions (Counter)
Number of identical execution requests that have been merged.
Labels
- group_id: Group (organization) ID associated with the request.
buildbuddy_remote_execution_hedged_actions (Counter)
Number of identicial execution request which were merged for which a hedged execution was run in the background.
Labels
- group_id: Group (organization) ID associated with the request.
buildbuddy_remote_execution_merged_actions_per_execution (Histogram)
Distribution of how many actions were submitted and merged against a single, canonical execution over the lifetime of that canonical execution.
Note that this metric is recorded once per merged-action, so distribution values are cumulative, or recorded n-times per canonical execution.
Labels
- group_id: Group (organization) ID associated with the request.
buildbuddy_remote_execution_merged_action_submit_time_offset_usec (Histogram)
The offset, in microseconds of wall-time, between the time when a merged action was submitted to the execution server and when the original action was submitted to the execution server.
Labels
- group_id: Group (organization) ID associated with the request.
Examples
# Rate of merged actions by group.
sum(rate(buildbuddy_remote_execution_merged_actions[1m])) by (group_id)
buildbuddy_remote_execution_queue_length (Gauge)
Number of actions currently waiting in the executor queue.
Labels
- group_id: Group (organization) ID associated with the request.
Examples
# Median queue length across all executors
quantile(0.5, buildbuddy_remote_execution_queue_length)
buildbuddy_remote_execution_tasks_executing (Gauge)
Number of tasks currently being executed by the executor.
Labels
- stage: Executed action stage. Action execution is split into stages corresponding to the timestamps defined in
ExecutedActionMetadata:queued,input_fetch,execution, andoutput_upload. An additional stage,worker, includes all stages during which a worker is handling the action, which is all stages except thequeuedstage.
Examples
# Fraction of idle executors
count_values(0, buildbuddy_remote_execution_tasks_executing)
/
count(buildbuddy_remote_execution_tasks_executing)
buildbuddy_remote_execution_assigned_ram_bytes (Gauge)
Estimated RAM on the executor that is currently allocated for task execution, in bytes.
buildbuddy_remote_execution_assigned_and_queued_estimated_ram_bytes (Gauge)
Estimated RAM on the executor that is currently allocated for queued or executing tasks, in bytes.
Note that this is a fuzzy estimate because there's no guarantee that tasks queued on a machine will be handled by that machine.
buildbuddy_remote_execution_assignable_ram_bytes (Gauge)
Maximum total RAM that can be allocated for task execution, in bytes.
buildbuddy_remote_execution_assignable_disk_bytes (Gauge)
Maximum total workspace disk that can be allocated for task execution, in bytes.
buildbuddy_remote_execution_assigned_milli_cpu (Gauge)
Estimated CPU time on the executor that is currently allocated for task execution, in milliCPU (CPU-milliseconds per second).
buildbuddy_remote_execution_assigned_and_queued_estimated_milli_cpu (Gauge)
Estimated CPU time on the executor that is currently allocated for queued or executing tasks, in milliCPU (CPU-milliseconds per second).
Note that this is a fuzzy estimate because there's no guarantee that tasks queued on a machine will be handled by that machine.
buildbuddy_remote_execution_assignable_milli_cpu (Gauge)
Maximum total CPU time on the executor that can be allocated for task execution, in milliCPU (CPU-milliseconds per second).
buildbuddy_remote_execution_assigned_custom_resources (Gauge)
Custom resources on the executor currently allocated for task execution.
Custom resources are dimensionless values configured via executor.custom_resources.
Labels
- resource_name: Name of a custom resource configured on an executor.
buildbuddy_remote_execution_assignable_custom_resources (Gauge)
Maximum custom resources that can be allocated for task execution.
Custom resources are dimensionless values configured via executor.custom_resources.
Labels
- resource_name: Name of a custom resource configured on an executor.
buildbuddy_remote_execution_file_download_count (Histogram)
Number of files downloaded during remote execution.
buildbuddy_remote_execution_file_download_size_bytes (Histogram)
Total number of bytes downloaded during remote execution.
buildbuddy_remote_execution_file_download_duration_usec (Histogram)
Per-file download duration during remote execution, in microseconds.
buildbuddy_remote_execution_file_upload_count (Histogram)
Number of files uploaded during remote execution.
buildbuddy_remote_execution_file_upload_size_bytes (Histogram)
Total number of bytes uploaded during remote execution.
buildbuddy_remote_execution_skipped_output_bytes (Counter)
Total number of output bytes that weren't uploaded after remote execution.
buildbuddy_remote_execution_file_upload_duration_usec (Histogram)
Per-file upload duration during remote execution, in microseconds.
buildbuddy_remote_execution_networking_command_duration_usec (Histogram)
Duration of networking commands, in microseconds.
Labels
- command: Command being run. Specific arguments to the command are omitted to reduce metric cardinality.
buildbuddy_remote_execution_networking_command_cpu_usage_usec (Histogram)
CPU usage of networking commands, in CPU-microseconds.
Labels
- command: Command being run. Specific arguments to the command are omitted to reduce metric cardinality.
buildbuddy_remote_execution_vfs_cas_files_count (Counter)
Total number of CAS files in VFS filesystems.
buildbuddy_remote_execution_vfs_cas_files_accessed_count (Counter)
Number of CAS files in VFS filesystems that were accessed by the action.
buildbuddy_remote_execution_vfs_cas_files_size_bytes (Counter)
Total size of CAS files in VFS filesystems.
buildbuddy_remote_execution_vfs_cas_files_accessed_bytes (Counter)
Size of CAS files in VFS filesystems that were accessed by the action.
buildbuddy_remote_runner_requests (Counter)
Number of remote runner executions initiated.
Labels
- group_id: Group (organization) ID associated with the request.
- op:
- stage: Generic label to describe the stage the metric is capturing
- os: OS associated with the request.
- arch: CPU architecture associated with the request.
- self_hosted: Whether the request is targeting self-hosted executors.
buildbuddy_firecracker_stage_duration_usec (Histogram)
The total duration of each firecracker stage, in microseconds.
Labels
- stage: Generic label to describe the stage the metric is capturing
Stage label values
- "init": Time for the VM to start up (either a new VM or from a snapshot)
- "exec": Time to run the command inside the container
- "task_lifecycle": Time from when the task if first assigned to the VM (beginning of init) to after it's finished execution. This roughly represents what a customer will wait for the task to complete after it's been scheduled to a firecracker runner
- "pause": Time to pause the VM, save a snapshot, and cleanup resources
Examples
# P95 workflow lifecycle duration in the past 5 minutes, grouped by group_id
histogram_quantile(
0.95,
sum by(le, group_id) (
rate(buildbuddy_firecracker_stage_duration_usec_bucket{job="executor-workflows", stage="task_lifecycle"}[5m])
)
)
buildbuddy_firecracker_snapshot_save_workloads_executing (Gauge)
Number of snapshot save workloads currently being executed by the executor.
Labels
- stage: Generic label to describe the stage the metric is capturing
buildbuddy_firecracker_snapshot_source_count (Counter)
The number of snapshot manifests fetched from the local vs remote cache.
Labels
- chunk_source: For chunked snapshot files, describes the initialization source of the chunk (Ex.
remote_cacheorlocal_filecache)
buildbuddy_firecracker_exec_dial_duration_usec (Histogram)
Time taken to dial the VM guest execution server after it has been started or resumed, in microseconds.
Labels
- created_from_snapshot: CreatedFromSnapshot indicates if a firecracker execution used a snapshot.
buildbuddy_firecracker_execution_error_count (Counter)
Count of Firecracker task execution errors, labeled by reason.
Labels
- reason: Reason for a Firecracker task execution failure.
buildbuddy_firecracker_snapshot_remote_cache_upload_size_bytes (Counter)
After a copy-on-write snapshot has been used, the total count of compressed bytes written to the cache (i.e.
will be 0 if the artifact is already cached).
Labels
- file_name: Name of a file.
buildbuddy_firecracker_cow_snapshot_dirty_chunk_ratio (Histogram)
After a copy-on-write snapshot has been used, the ratio of dirty/total chunks.
Labels
- file_name: Name of a file.
Examples
# To view how many elements fall into each bucket
# Visualize with the Bar Gauge type
# Legend: {{le}}
# Format: Heatmap
sum(increase(buildbuddy_firecracker_cow_snapshot_dirty_chunk_ratio_bucket[5m])) by(le)
buildbuddy_firecracker_cow_snapshot_dirty_bytes (Counter)
After a copy-on-write snapshot has been used, the total count of bytes dirtied.
Labels
- file_name: Name of a file.
buildbuddy_firecracker_cow_snapshot_bytes_read (Counter)
After a copy-on-write snapshot has been used, the number of bytes read from each source.
Labels
- file_name: Name of a file.
- chunk_source: For chunked snapshot files, describes the initialization source of the chunk (Ex.
remote_cacheorlocal_filecache)
buildbuddy_firecracker_cow_snapshot_chunk_source_ratio (Histogram)
After a copy-on-write snapshot has been used, the percentage of chunks that were initialized by the given source.
Labels
- file_name: Name of a file.
- chunk_source: For chunked snapshot files, describes the initialization source of the chunk (Ex.
remote_cacheorlocal_filecache)
buildbuddy_firecracker_cow_snapshot_memory_mapped_bytes (Gauge)
Total number of bytes currently memory-mapped.
Labels
- file_name: Name of a file.
buildbuddy_firecracker_cow_snapshot_page_fault_total_duration_usec (Histogram)
For a snapshotted VM, total time spent fulfilling page faults.
Labels
- stage: Generic label to describe the stage the metric is capturing
buildbuddy_firecracker_cow_snapshot_chunk_operation_duration_usec (Histogram)
For a COW snapshot, cumulative time spent on an operation type.
Labels
- file_name: Name of a file.
- name: The name used to identify the type of an unexpected event.
- stage: Generic label to describe the stage the metric is capturing
buildbuddy_firecracker_cow_bytes_read (Counter)
Total number of bytes read from COW chunked files.
Labels
- file_name: Name of a file.
buildbuddy_firecracker_cow_bytes_written (Counter)
Total number of bytes written to COW chunked files.
Labels
- file_name: Name of a file.
buildbuddy_firecracker_workspace_conversion_disk_write_ops (Counter)
Total number of disk write operations performed converting Firecracker action workspaces to/from ext4 images.
Labels
- command: Command being run. Specific arguments to the command are omitted to reduce metric cardinality.
buildbuddy_remote_execution_max_recyclable_resource_usage_event (Counter)
Counter for firecracker runners that reach max disk/memory usage and won't get recycled.
Labels
- group_id: Group (organization) ID associated with the request.
- name: The name used to identify the type of an unexpected event.
- recycled_runner_status: For remote execution runners, describes the recycling status (Ex. 'clean' if the runner is not recycled or 'recycled')
buildbuddy_remote_execution_recycle_runner_requests (Counter)
Number of execution requests with runner recycling enabled (via the platform property recycle-runner=true).
Labels
- status: Status of the recycle runner request:
hitif the executor assigned a recycled runner to the action;missotherwise.
buildbuddy_remote_execution_runner_pool_count (Gauge)
Number of command runners that are currently pooled (and available for recycling).
buildbuddy_remote_execution_runner_pool_evictions (Counter)
Number of command runners removed from the pool to make room for other runners.
buildbuddy_remote_execution_runner_pool_failed_recycle_attempts (Counter)
Number of failed attempts to add runners to the pool.
Labels
- reason: Reason for a runner not being added to the runner pool.
buildbuddy_remote_execution_runner_pool_memory_usage_bytes (Gauge)
Total memory usage of pooled command runners, in bytes.
Currently only supported for Docker-based executors.
buildbuddy_remote_execution_runner_pool_disk_usage_bytes (Gauge)
Total disk usage of pooled command runners, in bytes.
buildbuddy_remote_execution_file_cache_requests (Counter)
Number of local executor file cache requests.
Labels
- status: Status of the file cache request:
hitif found in cache,missotherwise.
buildbuddy_remote_execution_file_cache_op_latency_usec (Histogram)
Latency of individual file cache operations.
Labels
- op:
buildbuddy_remote_execution_file_cache_last_eviction_age_usec (Gauge)
Age of the last entry evicted from the executor's local file cache (relative to when it was added to the cache), in microseconds.
buildbuddy_remote_execution_file_cache_added_file_size_bytes (Histogram)
Size of artifacts added to the file cache, in bytes.
buildbuddy_remote_execution_file_cache_added_file_bytes_count (Counter)
Total number of bytes written to the filecache by groupid.
Labels
- group_id: Group (organization) ID associated with the request.
Blobstore metrics
"Blobstore" refers to the backing storage that BuildBuddy uses to store objects in the cache, as well as certain pieces of temporary data (such as invocation events while an invocation is in progress).
buildbuddy_blobstore_read_count (Counter)
Number of files read from the blobstore.
Labels
- status: Status code as defined by grpc/codes. This is a numeric value; any non-zero code indicates an error.
- blobstore_type:
gcs(Google Cloud Storage),aws_s3, ordisk.
buildbuddy_blobstore_read_size_bytes (Histogram)
Number of bytes read from the blobstore per file.
Labels
- blobstore_type:
gcs(Google Cloud Storage),aws_s3, ordisk.
Bytes downloaded per second
sum(rate(buildbuddy_blobstore_read_size_bytes[5m]))
### **`buildbuddy_blobstore_read_duration_usec`** (Histogram)
Duration per blobstore file read, in **microseconds**.
#### Labels
- **blobstore_type**: `gcs` (Google Cloud Storage), `aws_s3`, or `disk`.
### **`buildbuddy_blobstore_write_count`** (Counter)
Number of files written to the blobstore.
#### Labels
- **status**: Status code as defined by [grpc/codes](https://godoc.org/google.golang.org/grpc/codes#Code). This is a numeric value; any non-zero code indicates an error.
- **blobstore_type**: `gcs` (Google Cloud Storage), `aws_s3`, or `disk`.
# Bytes uploaded per second
sum(rate(buildbuddy_blobstore_write_size_bytes[5m]))
buildbuddy_blobstore_write_size_bytes (Histogram)
Number of bytes written to the blobstore per file.
Labels
- blobstore_type:
gcs(Google Cloud Storage),aws_s3, ordisk.
buildbuddy_blobstore_write_duration_usec (Histogram)
Duration per blobstore file write, in microseconds.
Labels
- blobstore_type:
gcs(Google Cloud Storage),aws_s3, ordisk.
buildbuddy_blobstore_clone_count (Counter)
Number of files cloned (copied storage-side, without the bytes leaving the blobstore).
Labels
- status: Status code as defined by grpc/codes. This is a numeric value; any non-zero code indicates an error.
- blobstore_type:
gcs(Google Cloud Storage),aws_s3, ordisk.
Bytes cloned per second (these bytes never leave the blobstore)
sum(rate(buildbuddy_blobstore_clone_size_bytes[5m]))
### **`buildbuddy_blobstore_clone_size_bytes`** (Histogram)
Number of bytes cloned per file.
#### Labels
- **blobstore_type**: `gcs` (Google Cloud Storage), `aws_s3`, or `disk`.
### **`buildbuddy_blobstore_clone_duration_usec`** (Histogram)
Duration per blobstore file clone, in **microseconds**.
#### Labels
- **blobstore_type**: `gcs` (Google Cloud Storage), `aws_s3`, or `disk`.
### **`buildbuddy_blobstore_delete_count`** (Counter)
Number of files deleted from the blobstore.
#### Labels
- **status**: Status code as defined by [grpc/codes](https://godoc.org/google.golang.org/grpc/codes#Code). This is a numeric value; any non-zero code indicates an error.
- **blobstore_type**: `gcs` (Google Cloud Storage), `aws_s3`, or `disk`.
### **`buildbuddy_blobstore_delete_duration_usec`** (Histogram)
Delete duration per blobstore file deletion, in **microseconds**.
#### Labels
- **blobstore_type**: `gcs` (Google Cloud Storage), `aws_s3`, or `disk`.
### **`buildbuddy_blobstore_exists_count`** (Counter)
Number of existance checks in the blobstore.
#### Labels
- **status**: Status code as defined by [grpc/codes](https://godoc.org/google.golang.org/grpc/codes#Code). This is a numeric value; any non-zero code indicates an error.
- **blobstore_type**: `gcs` (Google Cloud Storage), `aws_s3`, or `disk`.
### **`buildbuddy_blobstore_exists_duration_usec`** (Histogram)
CheckExists duration, in **microseconds**.
#### Labels
- **blobstore_type**: `gcs` (Google Cloud Storage), `aws_s3`, or `disk`.
### **`buildbuddy_invocation_log_bytes_written`** (Counter)
Number of invocation log bytes uploaded, either via the build event stream or the WriteEventLog API (stdout+stderr).
#### Labels
- **name**: The name used to identify the type of an unexpected event.
- **group_id**: Group (organization) ID associated with the request.
# SQL metrics
The following metrics are for monitoring the SQL database configured
for BuildBuddy.
If you'd like to see an up-to-date catalog of what BuildBuddy stores in
its SQL database, see the table definitions [here](https://github.com/buildbuddy-io/buildbuddy/blob/master/server/tables/tables.go).
## Query / error rate metrics
### **`buildbuddy_sql_query_count`** (Counter)
Number of SQL queries executed.
#### Labels
- **sql_query_template**: SQL query before substituting template parameters.
#### Examples
```promql
# SQL queries per second (by query template).
sum by (sql_query_template) (rate(buildbuddy_sql_query_count[5m]))
buildbuddy_sql_query_duration_usec (Histogram)
SQL query duration, in microseconds.
Labels
- sql_query_template: SQL query before substituting template parameters.
Examples
# Median SQL query duration
histogram_quantile(
0.5,
sum(rate(buildbuddy_sql_query_duration_usec_bucket[5m])) by (le)
)
buildbuddy_sql_error_count (Counter)
Number of SQL queries that resulted in an error.
Labels
- sql_query_template: SQL query before substituting template parameters.
Examples
# SQL error rate
sum(rate(buildbuddy_sql_error_count[5m]))
/
sum(rate(buildbuddy_sql_query_count[5m]))
database/sql metrics
The following metrics directly expose
DBStats from the
database/sql Go package.
buildbuddy_sql_max_open_connections (Gauge)
Maximum number of open connections to the database.
Labels
- sql_db_role: SQL DB replica role:
primaryfor read+write replicas, orread_replicafor read-only DB replicas.
buildbuddy_sql_open_connections (Gauge)
The number of established connections to the database.
Labels
- connection_status: Status of the database connection:
in_useoridle - sql_db_role: SQL DB replica role:
primaryfor read+write replicas, orread_replicafor read-only DB replicas.
buildbuddy_sql_wait_count (Counter)
The total number of connections waited for.
Labels
- sql_db_role: SQL DB replica role:
primaryfor read+write replicas, orread_replicafor read-only DB replicas.
buildbuddy_sql_wait_duration_usec (Counter)
The total time blocked waiting for a new connection, in microseconds.
Labels
- sql_db_role: SQL DB replica role:
primaryfor read+write replicas, orread_replicafor read-only DB replicas.
buildbuddy_sql_max_idle_closed (Counter)
The total number of connections closed due to SetMaxIdleConns.
Labels
- sql_db_role: SQL DB replica role:
primaryfor read+write replicas, orread_replicafor read-only DB replicas.
buildbuddy_sql_max_idle_time_closed (Counter)
The total number of connections closed due to SetConnMaxIdleTime.
Labels
- sql_db_role: SQL DB replica role:
primaryfor read+write replicas, orread_replicafor read-only DB replicas.
buildbuddy_sql_max_lifetime_closed (Counter)
The total number of connections closed due to SetConnMaxLifetime.
Labels
- sql_db_role: SQL DB replica role:
primaryfor read+write replicas, orread_replicafor read-only DB replicas.
HTTP metrics
buildbuddy_http_request_count (Counter)
HTTP request count.
Labels
- route: HTTP route before substituting path parameters (
/invocation/:id,/settings, ...) - method: HTTP method:
GET,POST, ...
Examples
# Requests per second, by status code
sum by (code) (rate(buildbuddy_http_request_count[5m]))
# 5xx error ratio
sum(rate(buildbuddy_http_request_count{code=~"5.."}[5m]))
/
sum(rate(buildbuddy_http_request_count[5m]))
buildbuddy_http_request_handler_duration_usec (Histogram)
Time taken to handle each HTTP request in microseconds.
Labels
- route: HTTP route before substituting path parameters (
/invocation/:id,/settings, ...) - method: HTTP method:
GET,POST, ... - code: HTTP response code:
200,302,401,404,500, ...
Examples
# Median request duration for successfully processed (2xx) requests.
# Other status codes may be associated with early-exits and are
# likely to add too much noise.
histogram_quantile(
0.5,
sum by (le) (rate(buildbuddy_http_request_handler_duration_usec{code=~"2.."}[5m]))
)
buildbuddy_http_response_size_bytes (Histogram)
Response size of each HTTP response in bytes.
Labels
- route: HTTP route before substituting path parameters (
/invocation/:id,/settings, ...) - method: HTTP method:
GET,POST, ... - code: HTTP response code:
200,302,401,404,500, ...
Examples
# Median HTTP response size
histogram_quantile(
0.5,
sum by (le) (rate(buildbuddy_http_response_size_bytes[5m]))
)
buildbuddy_http_client_request_count (Counter)
HTTP outgoing request count.
Labels
- client_name:
- host: Host (without any port number) for outgoing HTTP requests
- method: HTTP method:
GET,POST, ...
buildbuddy_http_client_response_size_bytes (Histogram)
Response size of response for each HTTP client request in bytes.
Labels
- client_name:
- host: Host (without any port number) for outgoing HTTP requests
- method: HTTP method:
GET,POST, ... - code: HTTP response code:
200,302,401,404,500, ...
Redis client metrics
buildbuddy_redis_client_dial_errors (Counter)
The number of errors when dialing redis.
Internal metrics
These metrics are for monitoring lower-level subsystems of BuildBuddy.
Build event handler
The build event handler logs all build events uploaded to BuildBuddy as part of the Build Event Protocol.
buildbuddy_build_event_handler_duration_usec (Histogram)
The time spent handling each build event in microseconds.
Labels
- status: Status code as defined by grpc/codes. This is a numeric value; any non-zero code indicates an error.
In-memory LRU caches
These metrics are reported by in-memory LRU caches (see server/util/lru) that are configured with a name.
buildbuddy_lru_lookup_count (Counter)
Number of LRU cache lookups (Get or Contains calls), by operation and hit/miss status.
Labels
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
- op: LRU cache operation:
getorcontains. - cache_status: Cache lookup result - One of: - "hit" - "miss" - "partial" (for batched RPCs where part of a request was cached) - Or "uncacheable" (for e.g. encrypted resources)
buildbuddy_lru_last_eviction_age_usec (Gauge)
Age of the entry most recently automatically evicted from an LRU cache (time since the entry was added or last updated), in microseconds.
Labels
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
- eviction_reason: The reason an entry was automatically evicted from an LRU cache:
size(evicted because LRU reached max size) orttl(expired).
Usage tracker
buildbuddy_usage_tracker_missing_collection_counts_count (Counter)
The number of times the usage tracking system was missing counts for a particular collection period.
This may happen if there are transient redis errors.
Webhooks
Webhooks are HTTP endpoints exposed by BuildBuddy server which allow it to respond to repository events. These URLs are created as part of BuildBuddy workflows.
buildbuddy_webhook_handler_workflows_started (Counter)
The number of workflows triggered by the webhook handler.
Labels
- event: Type of event sent to BuildBuddy's webhook handler:
pushorpull_request.
Cache
"Cache" refers to the cache backend(s) that BuildBuddy uses to accelerate file IO operations, which are common in different subsystems such as the remote cache and the fetch server (for downloading invocation artifacts).
BuildBuddy can be configured to use multiple layers of caching (an in-memory layer, coupled with a cloud storage layer).
get metrics
get metrics track non-streamed cache reads (all data is fetched
from the cache in a single request).
buildbuddy_cache_get_count (Counter)
Number of cache get requests.
Labels
- status: Status code as defined by grpc/codes. This is a numeric value; any non-zero code indicates an error.
- tier: Cache tier:
memoryorcloud. This label can be used to write Prometheus queries that don't break if the cache backend is swapped out for a different backend. - backend: Cache backend:
gcs(Google Cloud Storage),aws_s3, orredis.
buildbuddy_cache_get_duration_usec (Histogram)
The time spent retrieving each entry from the cache, in microseconds.
This is recorded only for successful gets.
Labels
- tier: Cache tier:
memoryorcloud. This label can be used to write Prometheus queries that don't break if the cache backend is swapped out for a different backend. - backend: Cache backend:
gcs(Google Cloud Storage),aws_s3, orredis.
buildbuddy_cache_get_size_bytes (Histogram)
Size of each entry retrieved from the cache, in bytes.
This is recorded only for successful gets.
Labels
- tier: Cache tier:
memoryorcloud. This label can be used to write Prometheus queries that don't break if the cache backend is swapped out for a different backend. - backend: Cache backend:
gcs(Google Cloud Storage),aws_s3, orredis.
read metrics
read metrics track streamed cache reads.
buildbuddy_cache_read_count (Counter)
Number of streamed cache reads started.
This is incremented once for each started stream, not for each chunk in the stream.
Labels
- status: Status code as defined by grpc/codes. This is a numeric value; any non-zero code indicates an error.
- tier: Cache tier:
memoryorcloud. This label can be used to write Prometheus queries that don't break if the cache backend is swapped out for a different backend. - backend: Cache backend:
gcs(Google Cloud Storage),aws_s3, orredis.
buildbuddy_cache_read_duration_usec (Histogram)
The total time spent for each read stream, in microseconds.
This is recorded only for successful reads, and measures the entire read stream (not just individual chunks).
Labels
- tier: Cache tier:
memoryorcloud. This label can be used to write Prometheus queries that don't break if the cache backend is swapped out for a different backend. - backend: Cache backend:
gcs(Google Cloud Storage),aws_s3, orredis.
buildbuddy_cache_read_size_bytes (Histogram)
Total size of each entry retrieved from the cache via streaming, in bytes.
This is recorded only on success, and measures the entire stream (not just individual chunks).
Labels
- tier: Cache tier:
memoryorcloud. This label can be used to write Prometheus queries that don't break if the cache backend is swapped out for a different backend. - backend: Cache backend:
gcs(Google Cloud Storage),aws_s3, orredis.
set metrics
set metrics track non-streamed cache writes (all data is wrtiten
in a single request).
buildbuddy_cache_set_count (Counter)
Number of cache set requests.
Labels
- status: Status code as defined by grpc/codes. This is a numeric value; any non-zero code indicates an error.
- tier: Cache tier:
memoryorcloud. This label can be used to write Prometheus queries that don't break if the cache backend is swapped out for a different backend. - backend: Cache backend:
gcs(Google Cloud Storage),aws_s3, orredis.
buildbuddy_cache_set_duration_usec (Histogram)
The time spent writing each entry to the cache, in microseconds.
This is recorded only for successful sets.
Labels
- tier: Cache tier:
memoryorcloud. This label can be used to write Prometheus queries that don't break if the cache backend is swapped out for a different backend. - backend: Cache backend:
gcs(Google Cloud Storage),aws_s3, orredis.
buildbuddy_cache_set_size_bytes (Histogram)
Size of the value stored in each set operation, in bytes.
This is recorded only for successful sets.
Labels
- tier: Cache tier:
memoryorcloud. This label can be used to write Prometheus queries that don't break if the cache backend is swapped out for a different backend. - backend: Cache backend:
gcs(Google Cloud Storage),aws_s3, orredis.
buildbuddy_cache_set_retries (Histogram)
Number of retries required to fulfill the set request (an observed value of 0 means the transfer succeeded on the first try).
Labels
- tier: Cache tier:
memoryorcloud. This label can be used to write Prometheus queries that don't break if the cache backend is swapped out for a different backend. - backend: Cache backend:
gcs(Google Cloud Storage),aws_s3, orredis.
write metrics
write metrics track streamed cache writes.
buildbuddy_cache_write_count (Counter)
Number of streamed cache writes started.
This is incremented once for each started stream, not for each chunk in the stream.
Labels
- status: Status code as defined by grpc/codes. This is a numeric value; any non-zero code indicates an error.
- tier: Cache tier:
memoryorcloud. This label can be used to write Prometheus queries that don't break if the cache backend is swapped out for a different backend. - backend: Cache backend:
gcs(Google Cloud Storage),aws_s3, orredis.
buildbuddy_cache_write_duration_usec (Histogram)
The time spent for each streamed write to the cache, in microseconds.
This is recorded only on success, and measures the entire stream (not just individual chunks).
Labels
- tier: Cache tier:
memoryorcloud. This label can be used to write Prometheus queries that don't break if the cache backend is swapped out for a different backend. - backend: Cache backend:
gcs(Google Cloud Storage),aws_s3, orredis.
buildbuddy_cache_write_size_bytes (Histogram)
Size of each entry written to the cache via streaming, in bytes.
This is recorded only on success, and measures the entire stream (not just individual chunks).
Labels
- tier: Cache tier:
memoryorcloud. This label can be used to write Prometheus queries that don't break if the cache backend is swapped out for a different backend. - backend: Cache backend:
gcs(Google Cloud Storage),aws_s3, orredis.
buildbuddy_cache_write_retries (Histogram)
Number of retries required to write each chunk in the stream (an observed value of 0 means the transfer succeeded on the first try).
Labels
- tier: Cache tier:
memoryorcloud. This label can be used to write Prometheus queries that don't break if the cache backend is swapped out for a different backend. - backend: Cache backend:
gcs(Google Cloud Storage),aws_s3, orredis.
Other cache metrics
buildbuddy_cache_delete_count (Counter)
Number of deletes from the cache.
Labels
- status: Status code as defined by grpc/codes. This is a numeric value; any non-zero code indicates an error.
- tier: Cache tier:
memoryorcloud. This label can be used to write Prometheus queries that don't break if the cache backend is swapped out for a different backend. - backend: Cache backend:
gcs(Google Cloud Storage),aws_s3, orredis.
buildbuddy_cache_delete_duration_usec (Histogram)
Duration of each cache deletion, in microseconds.
Labels
- tier: Cache tier:
memoryorcloud. This label can be used to write Prometheus queries that don't break if the cache backend is swapped out for a different backend. - backend: Cache backend:
gcs(Google Cloud Storage),aws_s3, orredis.
buildbuddy_cache_contains_count (Counter)
Number of contains(key) requests made to the cache.
Labels
- status: Status code as defined by grpc/codes. This is a numeric value; any non-zero code indicates an error.
- tier: Cache tier:
memoryorcloud. This label can be used to write Prometheus queries that don't break if the cache backend is swapped out for a different backend. - backend: Cache backend:
gcs(Google Cloud Storage),aws_s3, orredis.
buildbuddy_cache_contains_duration_usec (Histogram)
Duration of each each contains(key) request, in microseconds.
Labels
- tier: Cache tier:
memoryorcloud. This label can be used to write Prometheus queries that don't break if the cache backend is swapped out for a different backend. - backend: Cache backend:
gcs(Google Cloud Storage),aws_s3, orredis.
buildbuddy_cache_contains_retry_count (Histogram)
Number of retries required to fulfill each contains(key) request to the cache (an observed value of 0 means the request succeeded on the first try).
Labels
- tier: Cache tier:
memoryorcloud. This label can be used to write Prometheus queries that don't break if the cache backend is swapped out for a different backend. - backend: Cache backend:
gcs(Google Cloud Storage),aws_s3, orredis.
buildbuddy_cache_batch_operator_enqueued_digests (Counter)
The number of digests enqueued in a batch operator, with the outcome of the enqueue operation.
Labels
- operator_name:
- group_id: Group (organization) ID associated with the request.
- status: There are a couple of places where we enqueue RPCs to be batched and sent asynchronously: the atime_updater and the hit_tracker_client. This label tracks the outcome of these enqueue operations. One of "enqueued", "duplicate", "dropped_too_many_updates", or "dropped_too_many_batches"
buildbuddy_cache_batch_operator_flushed_digests (Counter)
The number of digests flushed from a batch operator.
Labels
- operator_name:
- group_id: Group (organization) ID associated with the request.
- status: Status code as defined by grpc/codes. This is a numeric value; any non-zero code indicates an error.
buildbuddy_cache_proxy_secondary_sync_write_digests (Counter)
The number of digests written synchronously to a secondary cache.
Labels
- server_name: Describes the name of the server that handles a client request, such as "byte_stream_server" or "cas_server"
- status: Status code as defined by grpc/codes. This is a numeric value; any non-zero code indicates an error.
Misc metrics
buildbuddy_version (Gauge)
Binary version of the running instance.
Always reports a value of 1 similar to the up metric, but has a label containing the version.
Labels
- version: Binary version. Example:
v2.0.0. - commit: Binary git commit SHA. Example:
4bd7046417608d785094aa5ec7aa009a9ae53753
buildbuddy_unexpected_event (Counter)
Counter for unexpected events.
Labels
- name: The name used to identify the type of an unexpected event.
buildbuddy_health_check_status (Gauge)
Health check status.
Labels
- health_check_name: Name of service the health check is running for (Ex "distributed_cache" or "sql_primary").
buildbuddy_quota_quota_exceeded_count (Counter)
Total number of calls banned by quota server
Labels
- quota_namespace: The namespace that is protected by quota manager.
- quota_key: The key used for quota accounting, either a group ID or an IP address.
buildbuddy_quota_quota_key_empty_count (Counter)
Total number of calls with empty quota key
Labels
- quota_namespace: The namespace that is protected by quota manager.
buildbuddy_clickhouse_insert_count (Counter)
Num of rows inserted into ClickHouse
Labels
- clickhouse_table_name: The name of the table in Clickhouse
- status: Status of the Clickhouse operation:
ok,error.
buildbuddy_clickhouse_query_duration_usec (Histogram)
ClickHouse query duration, in microseconds.
Labels
- sql_query_template: SQL query before substituting template parameters.
buildbuddy_clickhouse_error_count (Counter)
Number of ClickHouse SQL queries that resulted in an error.
Labels
- sql_query_template: SQL query before substituting template parameters.
- status: Status code as defined by grpc/codes in human-readable format, such as "OK" or "NotFound".
buildbuddy_clickhouse_query_count (Counter)
Number of ClickHouse SQL queries executed.
Labels
- sql_query_template: SQL query before substituting template parameters.
buildbuddy_compressor_bytes_compressed (Counter)
The number of decompressed bytes passed into compressors
Labels
- compression: Describes the type of compression
buildbuddy_compressor_bytes_decompressed (Counter)
The number of decompressed bytes passed out of decompressors
Labels
- compression: Describes the type of compression
buildbuddy_pebble_compression_ratio (Histogram)
The aggregate compression ratio (compressed / decompressed bytes) for a stream of data
Labels
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
- compression: Describes the type of compression
buildbuddy_compressor_compressed_blob_size_write (Counter)
The number of compressed bytes in all blobs
Labels
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
- compression: Describes the type of compression
buildbuddy_compressor_decompressed_blob_size_write (Counter)
The number of decompressed bytes in all blobs
Labels
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
- compression: Describes the type of compression
Examples
# Histogram buckets with the count of elements in each compression ratio bucket
# Visualize with the Bar Gauge type
# Legend: {{le}}
# Format: Heatmap
sum(increase(buildbuddy_pebble_compression_ratio_bucket[5m])) by(le)
# Percentage of elements that increased in size when compressed (compression ratio > 1)
# Visualize with the Stat type
(sum(buildbuddy_pebble_compression_ratio_count) - sum(buildbuddy_pebble_compression_ratio_bucket{le="1.0"})) / sum(buildbuddy_pebble_compression_ratio_count)
buildbuddy_server_upload_size_bytes (Histogram)
Number of bytes uploaded to the server in each upload.
Use the _sum suffix to get the total uploaded bytes and the _count suffix to get the number of uploaded files.
Labels
- cache_type: Cache type:
actionfor action cache,casfor content-addressable storage. - server_name: Describes the name of the server that handles a client request, such as "byte_stream_server" or "cas_server"
- tracked: Whether or not billable usage was recorded for this request ("true", "false")
buildbuddy_server_uncompressed_upload_bytes_count (Counter)
Number of uncompressed bytes uploaded to the server.
Labels
- cache_type: Cache type:
actionfor action cache,casfor content-addressable storage. - server_name: Describes the name of the server that handles a client request, such as "byte_stream_server" or "cas_server"
- group_id: Group (organization) ID associated with the request.
- tracked: Whether or not billable usage was recorded for this request ("true", "false")
buildbuddy_server_download_size_bytes (Histogram)
Number of bytes downloaded from the server in each download.
Use the _sum suffix to get the total downloaded bytes and the _count suffix to get the number of downloaded files.
Labels
- cache_type: Cache type:
actionfor action cache,casfor content-addressable storage. - server_name: Describes the name of the server that handles a client request, such as "byte_stream_server" or "cas_server"
- group_id: Group (organization) ID associated with the request.
- tracked: Whether or not billable usage was recorded for this request ("true", "false")
buildbuddy_server_uncompressed_download_bytes_count (Counter)
Number of uncompressed bytes downloaded from the server.
Labels
- cache_type: Cache type:
actionfor action cache,casfor content-addressable storage. - server_name: Describes the name of the server that handles a client request, such as "byte_stream_server" or "cas_server"
- group_id: Group (organization) ID associated with the request.
- tracked: Whether or not billable usage was recorded for this request ("true", "false")
buildbuddy_server_digest_upload_size_bytes (Histogram)
Digest size uploaded to the server in each upload.
This does not always match the actual size uploaded to the server, if the client sends compressed bytes. Use the _sum suffix to get the total uploaded bytes and the _count suffix to get the number of uploaded files.
Labels
- cache_type: Cache type:
actionfor action cache,casfor content-addressable storage. - server_name: Describes the name of the server that handles a client request, such as "byte_stream_server" or "cas_server"
- tracked: Whether or not billable usage was recorded for this request ("true", "false")
buildbuddy_server_digest_download_size_bytes (Histogram)
Digest size downloaded from the server in each download.
This does not always match the actual size downloaded to the server, if the client requests compressed bytes. Use the _sum suffix to get the total downloaded bytes and the _count suffix to get the number of downloaded files.
Labels
- cache_type: Cache type:
actionfor action cache,casfor content-addressable storage. - server_name: Describes the name of the server that handles a client request, such as "byte_stream_server" or "cas_server"
- tracked: Whether or not billable usage was recorded for this request ("true", "false")
buildbuddy_logger_log_count (Counter)
The number of logs
Labels
- status: Status code as defined by grpc/codes in human-readable format, such as "OK" or "NotFound".
Raft cache metrics
buildbuddy_raft_partition_operations (Counter)
Number of operations on partitions
Labels
- partition_id: The ID of the disk cache partition this event applied to.
- op: The type of operation on a partition: "initialize", "soft-delte", "hard-delete".
- status: Status code as defined by grpc/codes in human-readable format, such as "OK" or "NotFound".
buildbuddy_raft_ranges (Gauge)
Number of raft ranges on each nodehost.
Labels
- node_host_id: The ID of a partition.
buildbuddy_raft_range_replica (Gauge)
Indicator (always 1) that the labeled nodehost holds a replica of the given range.
Use to verify replica placement across zones and partitions.
Labels
- range_id: The range ID of a raft region.
- node_host_id: The ID of a partition.
- partition_id: The ID of the disk cache partition this event applied to.
- zone: The availability zone of a node.
buildbuddy_raft_leases (Gauge)
Number of raft leases on each nodehost.
Labels
- range_id: The range ID of a raft region.
- node_host_id: The ID of a partition.
- partition_id: The ID of the disk cache partition this event applied to.
- zone: The availability zone of a node.
buildbuddy_raft_leaders (Gauge)
Number of raft leaders on each nodehost.
Labels
- range_id: The range ID of a raft region.
- node_host_id: The ID of a partition.
- partition_id: The ID of the disk cache partition this event applied to.
- zone: The availability zone of a node.
buildbuddy_raft_bytes (Gauge)
Size (in bytes) of each range.
Labels
- range_id: The range ID of a raft region.
- node_host_id: The ID of a partition.
- partition_id: The ID of the disk cache partition this event applied to.
- zone: The availability zone of a node.
buildbuddy_raft_proposals (Counter)
The total number of statemachine proposals on each range.
Labels
- range_id: The range ID of a raft region.
- node_host_id: The ID of a partition.
- partition_id: The ID of the disk cache partition this event applied to.
- zone: The availability zone of a node.
buildbuddy_raft_reads (Counter)
The total number of read requests served for each range.
Labels
- range_id: The range ID of a raft region.
- node_host_id: The ID of a partition.
- partition_id: The ID of the disk cache partition this event applied to.
- zone: The availability zone of a node.
buildbuddy_raft_splits (Counter)
The total number of splits per nodehost.
Labels
- node_host_id: The ID of a partition.
- status: Status code as defined by grpc/codes in human-readable format, such as "OK" or "NotFound".
buildbuddy_raft_moves (Counter)
The total number of moves per nodehost.
Labels
- node_host_id: The ID of a partition.
- move_type: The type of raft move
add, orremove. - status: Status code as defined by grpc/codes in human-readable format, such as "OK" or "NotFound".
buildbuddy_raft_rangecache_lookups (Counter)
The total number of rangecache lookups per nodehost.
Labels
- rangecache_event_type: Raft RangeCache event type:
hit,miss, orupdate.
buildbuddy_raft_split_duration_usec (Histogram)
The time spent splitting a range in microseconds.
Labels
- range_id: The range ID of a raft region.
buildbuddy_raft_replica_update_duration_usec (Histogram)
The time spent on replica.Update in microseconds.
Labels
- range_id: The range ID of a raft region.
buildbuddy_raft_eviction_errors (Counter)
The total number of eviction errors
buildbuddy_raft_listener_events_dropped (Counter)
The total number of dropped listener events
Labels
- listener_id: The ID of a raft listener
- listener_event: Raft Listener Event Type, such as "LeaderUpdated". These are events sent by dragonboat library.
buildbuddy_raft_store_events (Gauge)
Number of events in the queue
buildbuddy_raft_store_event_broadcast_dropped (Counter)
The total number of dropped events to broadcast
Labels
- event_broadcast_source: The source of the event broadcast.
- raft_event: Raft Event Type, such as "range-removed", "range-usage-updated"
buildbuddy_raft_store_event_listener_dropped (Counter)
The total number of dropped events in store
Labels
- listener_id: The ID of a raft listener
- raft_event: Raft Event Type, such as "range-removed", "range-usage-updated"
buildbuddy_raft_lease_action_count (Counter)
The total number of lease actions
Labels
- range_id: The range ID of a raft region.
- lease_action: The type of lease action
Acquire,Drop. - status: Status code as defined by grpc/codes in human-readable format, such as "OK" or "NotFound".
buildbuddy_raft_lease_action_duration_msec (Histogram)
The duration of a lease action
Labels
- lease_action: The type of lease action
Acquire,Drop.
buildbuddy_raft_zombie_cleanup_tasks (Gauge)
The total number of pending zombie cleanup tasks
buildbuddy_raft_zombie_cleanup (Counter)
The total number of zombie cleanups
Labels
- status: Status code as defined by grpc/codes in human-readable format, such as "OK" or "NotFound".
buildbuddy_raft_range_lock_duration_msec (Histogram)
The duration of the range lock in a client session
Labels
- range_id: The range ID of a raft region.
buildbuddy_raft_sender_range_lock_duration_msec (Histogram)
Time spent waiting for and holding the sender-side per-range propose lock, which serializes proposes across the full retry lifecycle of a single logical request to keep retries idempotent.
Labels
- range_id: The range ID of a raft region.
buildbuddy_raft_nodehost_method_errors (Counter)
The total number of nodehost method
Labels
- nodehost_method: The type of nodehost method: "SyncPropose", "SyncRead"
- dragonboat_error: The error returned by dragonboat library
buildbuddy_raft_nodehost_method_usec (Histogram)
The duration of a nodehost method
Labels
- nodehost_method: The type of nodehost method: "SyncPropose", "SyncRead"
- range_id: The range ID of a raft region.
buildbuddy_raft_batch_atime_update_usec (Histogram)
The duration of the batch request to update atime
buildbuddy_raft_batch_delete_usec (Histogram)
The duration of the batch request to delete
buildbuddy_raft_driver_actions (Counter)
The total number of driver action
Labels
- driver_action: Raft Driver Action
- driver_requeue_type: Raft Driver Requeue Type
- range_id: The range ID of a raft region.
buildbuddy_raft_intermediate_replicas_count (Gauge)
The number of intermediate replicas
Labels
- replica_state: Raft Replica State: staging, removing
- range_id: The range ID of a raft region.
buildbuddy_raft_txn_record_process_count (Counter)
The total number of driver action
Labels
- txn_record_process_status: The status of processing txn record: "success" or "failure"
buildbuddy_raft_eviction_samples_chan_size (Gauge)
Num of items in eviction samples chan
Labels
- partition_id: The ID of the disk cache partition this event applied to.
buildbuddy_raft_eviction_gcs_chan_size (Gauge)
Num of items in gcs eviction chan
Labels
- partition_id: The ID of the disk cache partition this event applied to.
buildbuddy_raft_gcs_delete_dropped (Counter)
The total number of dropped gcs deletes
Labels
- partition_id: The ID of the disk cache partition this event applied to.
buildbuddy_raft_atime_update_gcs_count (Counter)
Count of atime updates to GCS.
Labels
- status: Status code as defined by grpc/codes in human-readable format, such as "OK" or "NotFound".
buildbuddy_raft_gcs_eviction_count (Counter)
Count of evictions from GCS.
Labels
- partition_id: The ID of the disk cache partition this event applied to.
- status: Status code as defined by grpc/codes in human-readable format, such as "OK" or "NotFound".
buildbuddy_auth_api_key_lookup_count (Counter)
Total number of API key lookups.
Labels
- status: Whether or not the API Key lookup hit the in memory cache or not: "cache_hit", "cache_miss" or "invalid_key".
buildbuddy_auth_ip_rules_check_latency_usec (Histogram)
Latency of IP authorization checks.
Labels
- status: Status code as defined by grpc/codes in human-readable format, such as "OK" or "NotFound".
buildbuddy_auth_jwt_verification_count (Counter)
Total number of JWT verifications by signing-method and outcome.
Labels
- method: Signing algorithm used (JWT alg), such as "HS256" or "ES256".
- status: Status code as defined by grpc/codes in human-readable format, such as "OK" or "NotFound".
buildbuddy_encryption_key_refresh_count (Counter)
Total number of encryption key refresh attempts.
buildbuddy_encryption_key_refresh_failure_count (Counter)
Total number of unsuccessful encryption key refresh attempts.
buildbuddy_encryption_encrypted_block_count (Counter)
Total number of blocks encrypted.
buildbuddy_encryption_encrypted_blob_count (Counter)
Total number of blobs encrypted.
buildbuddy_encryption_decrypted_block_count (Counter)
Total number of blocks decrypted.
buildbuddy_encryption_decrypted_blob_count (Counter)
Total number of blobs decrypted.
buildbuddy_encryption_decryption_error_count (Counter)
Total number of decryption errors.
buildbuddy_encryption_key_last_encryption_age_msec (Histogram)
Age of encrypted keys (i.e.
how long it has been since the keys were re-encrypted).
buildbuddy_remote_cache_pebble_cache_atime_update_count (Counter)
Count of processed atime updates.
Labels
- partition_id: The ID of the disk cache partition this event applied to.
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
buildbuddy_remote_cache_pebble_cache_atime_update_gcs_error_count (Counter)
Count of atime update errors from GCS.
Labels
- partition_id: The ID of the disk cache partition this event applied to.
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
buildbuddy_remote_cache_pebble_cache_atime_delta_when_read (Histogram)
Previous atime of items in the cache when they are read, in msec
Labels
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
buildbuddy_remote_cache_pebble_cache_eviction_samples_chan_size (Gauge)
Num of items in eviction samples chan
Labels
- partition_id: The ID of the disk cache partition this event applied to.
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
buildbuddy_remote_cache_pebble_cache_eviction_resample_latency_usec (Histogram)
Latency of resampling during a single eviction iteration.
Labels
- partition_id: The ID of the disk cache partition this event applied to.
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
buildbuddy_remote_cache_pebble_cache_eviction_evict_latency_usec (Histogram)
Latency of evicting a single key.
Labels
- partition_id: The ID of the disk cache partition this event applied to.
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
buildbuddy_remote_cache_pebble_cache_eviction_samples (Counter)
Number of samples produced, by status
Labels
- partition_id: The ID of the disk cache partition this event applied to.
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
- status: Status code as defined by grpc/codes. This is a numeric value; any non-zero code indicates an error.
buildbuddy_remote_cache_pebble_cache_pebble_compact_count (Counter)
Number of compactions performed by the underlying Pebble database.
Labels
- compaction_type: Pebble DB compaction type.
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
buildbuddy_remote_cache_pebble_cache_pebble_compact_estimated_debt_bytes (Gauge)
Estimated number of bytes that need to be compacted for the LMS to reach a stable state.
Labels
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
buildbuddy_remote_cache_pebble_cache_pebble_compact_in_progress_bytes (Gauge)
Number of bytes present in sstables being written by in-progress compactions.
Labels
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
buildbuddy_remote_cache_pebble_cache_pebble_compact_in_progress (Gauge)
Number of compactions that are in-progress
Labels
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
buildbuddy_remote_cache_pebble_cache_pebble_compact_marked_files (Gauge)
Count of files that are marked for compaction.
Labels
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
buildbuddy_remote_cache_pebble_cache_pebble_level_sublevels (Gauge)
Number of sublevels within the level.
Labels
- level: Pebble DB level number.
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
buildbuddy_remote_cache_pebble_cache_pebble_level_num_files (Gauge)
The total number of files in the level.
Labels
- level: Pebble DB level number.
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
buildbuddy_remote_cache_pebble_cache_pebble_level_size_bytes (Gauge)
The total size in bytes of the files in the level.
Labels
- level: Pebble DB level number.
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
buildbuddy_remote_cache_pebble_cache_pebble_level_score (Gauge)
The level's compaction score.
Labels
- level: Pebble DB level number.
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
buildbuddy_remote_cache_pebble_cache_pebble_level_bytes_in_count (Counter)
The number of incoming bytes from other levels read during compactions.
This excludes bytes moved and bytes ingested. For L0 this is the bytes written to the WAL.
Labels
- level: Pebble DB level number.
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
buildbuddy_remote_cache_pebble_cache_pebble_level_bytes_ingested_count (Counter)
The number of bytes ingested.
Labels
- level: Pebble DB level number.
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
buildbuddy_remote_cache_pebble_cache_pebble_level_bytes_moved_count (Counter)
The number of bytes moved into the level by a move compaction.
Labels
- level: Pebble DB level number.
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
buildbuddy_remote_cache_pebble_cache_pebble_level_bytes_read_count (Counter)
The number of bytes read for compactions at the level.
This includes bytes read from other levels (BytesIn), as well as bytes read for the level.
Labels
- level: Pebble DB level number.
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
buildbuddy_remote_cache_pebble_cache_pebble_level_bytes_compacted_count (Counter)
The number of bytes written during compactions.
Labels
- level: Pebble DB level number.
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
buildbuddy_remote_cache_pebble_cache_pebble_level_bytes_flushed_count (Counter)
The number of bytes written during flushes.
Labels
- level: Pebble DB level number.
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
buildbuddy_remote_cache_pebble_cache_pebble_level_tables_compacted_count (Counter)
The number of sstables compacted to this level.
Labels
- level: Pebble DB level number.
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
buildbuddy_remote_cache_pebble_cache_pebble_level_tables_flushed_count (Counter)
The number of sstables flushed to this level.
Labels
- level: Pebble DB level number.
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
buildbuddy_remote_cache_pebble_cache_pebble_level_tables_ingested_count (Counter)
The number of sstables ingested into this level.
Labels
- level: Pebble DB level number.
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
buildbuddy_remote_cache_pebble_cache_pebble_level_tables_moved_count (Counter)
The number of sstables ingested into to this level.
Labels
- level: Pebble DB level number.
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
buildbuddy_remote_cache_pebble_cache_pebble_op_count (Counter)
The number of operations performed against the pebble database.
Labels
- pebble_id: Pebble DB ID
- pebble_op: Pebble DB operation type.
buildbuddy_remote_cache_pebble_cache_pebble_op_latency_usec (Histogram)
The latency of operations performed against the pebble database, in microseconds.
Labels
- pebble_id: Pebble DB ID
- pebble_op: Pebble DB operation type.
buildbuddy_remote_cache_pebble_cache_pebble_block_cache_size_bytes (Gauge)
The total size in pebble's block cache.
Labels
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
buildbuddy_remote_cache_pebble_cache_pebble_block_cache_requests_count (Counter)
The number of block cache requests by hit/miss status.
Labels
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
- cache_status: Cache lookup result - One of: - "hit" - "miss" - "partial" (for batched RPCs where part of a request was cached) - Or "uncacheable" (for e.g. encrypted resources)
buildbuddy_remote_cache_pebble_cache_pebble_table_cache_requests_count (Counter)
The number of table cache requests by hit/miss status.
Labels
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
- cache_status: Cache lookup result - One of: - "hit" - "miss" - "partial" (for batched RPCs where part of a request was cached) - Or "uncacheable" (for e.g. encrypted resources)
buildbuddy_remote_cache_pebble_cache_pebble_write_stall_count (Gauge)
The number of write stalls
Labels
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
buildbuddy_remote_cache_pebble_cache_pebble_write_stall_duration_usec (Histogram)
The duration of write stall in pebble, in microseconds.
Labels
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
buildbuddy_remote_cache_pebble_cache_zombie_table_count (Gauge)
The number of zombie tables
Labels
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
buildbuddy_remote_cache_pebble_cache_zombie_table_size_bytes (Gauge)
The size of zombie tables in bytes
Labels
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
buildbuddy_remote_cache_pebble_cache_find_missing_digest_count (Counter)
Count of digests within FindMissing requests.
Labels
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
buildbuddy_remote_cache_pebble_cache_find_missing_blob_status_count (Counter)
Count of blobs checked by FindMissing on this node, by present/absent status and originating purpose.
This is the per-node LOOKUP view: a blob retried across replicas by the distributed cache is counted on each node, so absents are inflated by replication vs the logical rate (see distributed_cache_find_missing_blob_status_count).
Labels
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
- purpose: FindMissing purpose: which internal code path originated a FindMissing lookup (see repb.FindMissingBlobsRequest.Purpose), e.g. "ATIME_UPDATE".
- status: Status code as defined by grpc/codes. This is a numeric value; any non-zero code indicates an error.
buildbuddy_remote_cache_pebble_cache_presence_cache_entry_count (Gauge)
Number of entries in the cache.
Labels
- cache_name: Cache name: Custom name to describe the cache, like "pebble-cache".
Podman metrics
buildbuddy_podman_image_pull_latency_msec (Histogram)
The latency of 'cold' podman pull requests per image, in milliseconds.
'Cold' means the image hasn't been pulled by this executor previously.
Labels
- container_image_tag: Container image tag.
Cache Proxy metrics
buildbuddy_proxy_action_cache_read_requests (Counter)
The number of ActionCache.GetActionResult requests served by a ActionCacheServerProxy by gRPC status, cache hit/miss status, request type, and result source.
Labels
- status: Status code as defined by grpc/codes. This is a numeric value; any non-zero code indicates an error.
- cache_status: Cache lookup result - One of: - "hit" - "miss" - "partial" (for batched RPCs where part of a request was cached) - Or "uncacheable" (for e.g. encrypted resources)
- proxy_request_type: Cache proxy request type. If "local_only", indicates the proxy should only use its local cache to fulfil the request. If "default", the proxy should fall back to the remote cache as the source of truth.
- result_source: Source used by a cache proxy request to produce a response.
buildbuddy_proxy_action_cache_write_requests (Counter)
The number of ActionCache.UpdateActionResult requests served by a ActionCacheServerProxy by gRPC status and cache hit/miss status.
Labels
- status: Status code as defined by grpc/codes. This is a numeric value; any non-zero code indicates an error.
- cache_status: Cache lookup result - One of: - "hit" - "miss" - "partial" (for batched RPCs where part of a request was cached) - Or "uncacheable" (for e.g. encrypted resources)
- proxy_request_type: Cache proxy request type. If "local_only", indicates the proxy should only use its local cache to fulfil the request. If "default", the proxy should fall back to the remote cache as the source of truth.
buildbuddy_proxy_action_cache_read_bytes (Counter)
The number of ActionCache.GetActionResult bytes served by a ActionCacheServerProxy by gRPC status, cache hit/miss status, request type, and result source.
Labels
- status: Status code as defined by grpc/codes. This is a numeric value; any non-zero code indicates an error.
- cache_status: Cache lookup result - One of: - "hit" - "miss" - "partial" (for batched RPCs where part of a request was cached) - Or "uncacheable" (for e.g. encrypted resources)
- proxy_request_type: Cache proxy request type. If "local_only", indicates the proxy should only use its local cache to fulfil the request. If "default", the proxy should fall back to the remote cache as the source of truth.
- result_source: Source used by a cache proxy request to produce a response.
buildbuddy_proxy_action_cache_write_bytes (Counter)
The number of ActionCache.UpdateActionResult bytes served by a ActionCacheServerProxy by gRPC status and cache hit/miss status.
Labels
- status: Status code as defined by grpc/codes. This is a numeric value; any non-zero code indicates an error.
- cache_status: Cache lookup result - One of: - "hit" - "miss" - "partial" (for batched RPCs where part of a request was cached) - Or "uncacheable" (for e.g. encrypted resources)
- proxy_request_type: Cache proxy request type. If "local_only", indicates the proxy should only use its local cache to fulfil the request. If "default", the proxy should fall back to the remote cache as the source of truth.
buildbuddy_proxy_byte_stream_read_requests (Counter)
The number of ByteStream.Read requests served by a ByteStreamServerProxy broken down by gRPC status and cache hit/miss status.
Labels
- status: Status code as defined by grpc/codes. This is a numeric value; any non-zero code indicates an error.
- cache_status: Cache lookup result - One of: - "hit" - "miss" - "partial" (for batched RPCs where part of a request was cached) - Or "uncacheable" (for e.g. encrypted resources)
- proxy_request_type: Cache proxy request type. If "local_only", indicates the proxy should only use its local cache to fulfil the request. If "default", the proxy should fall back to the remote cache as the source of truth.
- compression: Describes the type of compression
- chunked: Whether the request was handled using chunking
buildbuddy_proxy_byte_stream_write_requests (Counter)
The number of ByteStream.Write requests served by a ByteStreamServerProxy broken down by gRPC status and cache hit/miss status.
Labels
- status: Status code as defined by grpc/codes. This is a numeric value; any non-zero code indicates an error.
- cache_status: Cache lookup result - One of: - "hit" - "miss" - "partial" (for batched RPCs where part of a request was cached) - Or "uncacheable" (for e.g. encrypted resources)
- proxy_request_type: Cache proxy request type. If "local_only", indicates the proxy should only use its local cache to fulfil the request. If "default", the proxy should fall back to the remote cache as the source of truth.
- compression: Describes the type of compression
- chunked: Whether the request was handled using chunking
buildbuddy_proxy_byte_stream_read_bytes (Counter)
The number of bytes read by ByteStream.Read RPCs served by a ByteStreamServerProxy broken down by gRPC status and cache hit/miss status.
Note: this metric tracks bytes sent over the wire, which may be compressed.
Labels
- status: Status code as defined by grpc/codes. This is a numeric value; any non-zero code indicates an error.
- cache_status: Cache lookup result - One of: - "hit" - "miss" - "partial" (for batched RPCs where part of a request was cached) - Or "uncacheable" (for e.g. encrypted resources)
- proxy_request_type: Cache proxy request type. If "local_only", indicates the proxy should only use its local cache to fulfil the request. If "default", the proxy should fall back to the remote cache as the source of truth.
- compression: Describes the type of compression
- chunked: Whether the request was handled using chunking
- group_id: Group (organization) ID associated with the request.
buildbuddy_proxy_byte_stream_write_bytes (Counter)
The number of bytes written to a ByteStream.Write RPCs served by a ByteStreamServerProxy broken down by gRPC status and cache hit/miss status.
Note: this metric tracks bytes sent over the wire, which may be compressed.
Labels
- status: Status code as defined by grpc/codes. This is a numeric value; any non-zero code indicates an error.
- cache_status: Cache lookup result - One of: - "hit" - "miss" - "partial" (for batched RPCs where part of a request was cached) - Or "uncacheable" (for e.g. encrypted resources)
- proxy_request_type: Cache proxy request type. If "local_only", indicates the proxy should only use its local cache to fulfil the request. If "default", the proxy should fall back to the remote cache as the source of truth.
- compression: Describes the type of compression
- chunked: Whether the request was handled using chunking
buildbuddy_proxy_byte_stream_chunked_write_blob_bytes (Counter)
Original blob size in bytes for chunked writes.
Labels
- status: Status code as defined by grpc/codes. This is a numeric value; any non-zero code indicates an error.
- compression: Describes the type of compression
buildbuddy_proxy_byte_stream_chunked_write_chunk_bytes_total (Counter)
Total chunk bytes produced during chunked writes (sum of all chunk sizes).
Labels
- status: Status code as defined by grpc/codes. This is a numeric value; any non-zero code indicates an error.
- compression: Describes the type of compression
buildbuddy_proxy_byte_stream_chunked_write_chunk_bytes_deduped (Counter)
Chunk bytes that were deduplicated (already existed on remote) during chunked writes.
Labels
- status: Status code as defined by grpc/codes. This is a numeric value; any non-zero code indicates an error.
- compression: Describes the type of compression
buildbuddy_proxy_byte_stream_chunked_write_chunks_total (Counter)
Total number of chunks produced during chunked writes.
Labels
- status: Status code as defined by grpc/codes. This is a numeric value; any non-zero code indicates an error.
- compression: Describes the type of compression
- group_id: Group (organization) ID associated with the request.
buildbuddy_proxy_byte_stream_chunked_write_chunks_deduped (Counter)
Number of chunks that were deduplicated (already existed on remote) during chunked writes.
Labels
- status: Status code as defined by grpc/codes. This is a numeric value; any non-zero code indicates an error.
- compression: Describes the type of compression
- group_id: Group (organization) ID associated with the request.
buildbuddy_proxy_byte_stream_chunked_write_by_group_chunk_bytes_total (Counter)
Total chunk bytes produced during chunked writes by group ID (sum of all chunk sizes).
Labels
- status: Status code as defined by grpc/codes. This is a numeric value; any non-zero code indicates an error.
- compression: Describes the type of compression
- group_id: Group (organization) ID associated with the request.
buildbuddy_proxy_byte_stream_chunked_write_by_group_chunk_bytes_deduped (Counter)
Chunk bytes that were deduplicated (already existed on remote) during chunked writes by group ID.
Labels
- status: Status code as defined by grpc/codes. This is a numeric value; any non-zero code indicates an error.
- compression: Describes the type of compression
- group_id: Group (organization) ID associated with the request.
buildbuddy_proxy_byte_stream_chunked_write_by_action_mnemonic_chunk_bytes_total (Counter)
Total chunk bytes produced during chunked writes by action mnemonic (sum of all chunk sizes).
Labels
- status: Status code as defined by grpc/codes. This is a numeric value; any non-zero code indicates an error.
- compression: Describes the type of compression
- action_mnemonic: Short action name, such as "GoCompile".
buildbuddy_proxy_byte_stream_chunked_write_by_action_mnemonic_chunk_bytes_deduped (Counter)
Chunk bytes that were deduplicated (already existed on remote) during chunked writes by action mnemonic.
Labels
- status: Status code as defined by grpc/codes. This is a numeric value; any non-zero code indicates an error.
- compression: Describes the type of compression
- action_mnemonic: Short action name, such as "GoCompile".
buildbuddy_proxy_byte_stream_chunked_write_by_action_mnemonic_chunks_total (Counter)
Total number of chunks produced during chunked writes by action mnemonic.
Labels
- status: Status code as defined by grpc/codes. This is a numeric value; any non-zero code indicates an error.
- compression: Describes the type of compression
- action_mnemonic: Short action name, such as "GoCompile".
buildbuddy_proxy_byte_stream_chunked_write_by_action_mnemonic_chunks_deduped (Counter)
Number of chunks that were deduplicated (already existed on remote) during chunked writes by action mnemonic.
Labels
- status: Status code as defined by grpc/codes. This is a numeric value; any non-zero code indicates an error.
- compression: Describes the type of compression
- action_mnemonic: Short action name, such as "GoCompile".
buildbuddy_cache_client_chunked_upload_chunk_bytes_total (Counter)
Total uncompressed chunk bytes produced during chunked uploads.
Labels
- group_id: Group (organization) ID associated with the request.
buildbuddy_cache_client_chunked_upload_chunk_bytes_deduped (Counter)
Uncompressed chunk bytes deduplicated (already existed on remote) during chunked uploads.
Labels
- group_id: Group (organization) ID associated with the request.
buildbuddy_cache_client_chunked_upload_chunks_total (Counter)
Total number of chunks produced during chunked uploads.
Labels
- group_id: Group (organization) ID associated with the request.
buildbuddy_cache_client_chunked_upload_chunks_deduped (Counter)
Number of chunks deduplicated (already existed on remote) during chunked uploads.
Labels
- group_id: Group (organization) ID associated with the request.
buildbuddy_cache_client_chunked_upload_by_action_mnemonic_chunk_bytes_total (Counter)
Total uncompressed chunk bytes produced during chunked uploads by action mnemonic.
Labels
- action_mnemonic: Short action name, such as "GoCompile".
buildbuddy_cache_client_chunked_upload_by_action_mnemonic_chunk_bytes_deduped (Counter)
Uncompressed chunk bytes deduplicated (already existed on remote) during chunked uploads by action mnemonic.
Labels
- action_mnemonic: Short action name, such as "GoCompile".
buildbuddy_cache_client_chunked_upload_by_action_mnemonic_chunks_total (Counter)
Total number of chunks produced during chunked uploads by action mnemonic.
Labels
- action_mnemonic: Short action name, such as "GoCompile".
buildbuddy_cache_client_chunked_upload_by_action_mnemonic_chunks_deduped (Counter)
Number of chunks deduplicated (already existed on remote) during chunked uploads by action mnemonic.
Labels
- action_mnemonic: Short action name, such as "GoCompile".
buildbuddy_cache_client_chunked_download_chunk_bytes_total (Counter)
Total uncompressed chunk bytes read during chunked downloads.
buildbuddy_cache_client_chunked_download_chunk_bytes_local (Counter)
Uncompressed chunk bytes served from a local file during chunked downloads.
buildbuddy_cache_client_chunked_download_chunks_total (Counter)
Total number of chunks read during chunked downloads.
buildbuddy_cache_client_chunked_download_chunks_local (Counter)
Number of chunks served from a local file during chunked downloads.
buildbuddy_proxy_byte_stream_chunked_write_duration_usec (Histogram)
Total duration of chunked write operations, in microseconds.
Labels
- status: Status code as defined by grpc/codes. This is a numeric value; any non-zero code indicates an error.
- compression: Describes the type of compression
buildbuddy_proxy_byte_stream_chunked_write_chunking_duration_usec (Histogram)
Duration of the chunking phase (receive, decompress, chunk, compress, local write), in microseconds.
Labels
- status: Status code as defined by grpc/codes. This is a numeric value; any non-zero code indicates an error.
- compression: Describes the type of compression
buildbuddy_proxy_byte_stream_chunked_write_remote_duration_usec (Histogram)
Duration of the remote phase (FindMissingBlobs, upload, SpliceBlob), in microseconds.
Labels
- status: Status code as defined by grpc/codes. This is a numeric value; any non-zero code indicates an error.
- compression: Describes the type of compression
buildbuddy_proxy_byte_stream_chunked_write_upload_size_bytes (Histogram)
Compressed upload size for remote chunk uploads during chunked writes.
Labels
- status: Status code as defined by grpc/codes. This is a numeric value; any non-zero code indicates an error.
buildbuddy_proxy_byte_stream_chunked_read_requests (Counter)
Total number of successful chunked read requests.
Labels
- status: Status code as defined by grpc/codes. This is a numeric value; any non-zero code indicates an error.
- compression: Describes the type of compression
buildbuddy_proxy_byte_stream_chunked_read_blob_bytes (Counter)
Original blob size in bytes for chunked reads.
Labels
- status: Status code as defined by grpc/codes. This is a numeric value; any non-zero code indicates an error.
- compression: Describes the type of compression
buildbuddy_proxy_byte_stream_chunked_read_chunks_total (Counter)
Total number of chunks read during chunked reads.
Labels
- status: Status code as defined by grpc/codes. This is a numeric value; any non-zero code indicates an error.
- compression: Describes the type of compression
buildbuddy_proxy_byte_stream_chunked_read_chunks_local (Counter)
Number of chunks served from local cache during chunked reads.
Labels
- status: Status code as defined by grpc/codes. This is a numeric value; any non-zero code indicates an error.
- compression: Describes the type of compression
buildbuddy_proxy_byte_stream_chunked_read_chunks_remote (Counter)
Number of chunks fetched from remote during chunked reads.
Labels
- status: Status code as defined by grpc/codes. This is a numeric value; any non-zero code indicates an error.
- compression: Describes the type of compression
buildbuddy_proxy_byte_stream_chunked_read_bytes_local (Counter)
Bytes served from local cache to the client during chunked reads.
Labels
- status: Status code as defined by grpc/codes. This is a numeric value; any non-zero code indicates an error.
- compression: Describes the type of compression
buildbuddy_proxy_byte_stream_chunked_read_bytes_remote (Counter)
Bytes fetched from remote and served to the client during chunked reads.
Labels
- status: Status code as defined by grpc/codes. This is a numeric value; any non-zero code indicates an error.
- compression: Describes the type of compression
buildbuddy_proxy_byte_stream_chunked_read_fast_path_attempts (Counter)
Number of chunked reads where the CDC read fast path was attempted, by outcome.
Labels
- outcome: Outcome of a cache proxy fast path attempt.
buildbuddy_proxy_byte_stream_chunked_read_local_manifest_store_attempts (Counter)
Number of attempts to store a CDC manifest in the proxy local cache after a successful remote SplitBlob.
Labels
- status: Status code as defined by grpc/codes in human-readable format, such as "OK" or "NotFound".
buildbuddy_proxy_byte_stream_chunked_read_local_write_back_failures (Counter)
Number of failures while writing remotely fetched chunks back to the local cache during chunked reads.
Labels
- status: Status code as defined by grpc/codes in human-readable format, such as "OK" or "NotFound".
buildbuddy_proxy_byte_stream_chunked_read_failures (Counter)
Number of chunked read attempts that failed, by failure reason.
Labels
- reason:
- status: Status code as defined by grpc/codes in human-readable format, such as "OK" or "NotFound".
buildbuddy_remote_cache_byte_stream_chunked_read_failures (Counter)
Number of chunked read attempts that failed on the backend, by failure reason.
Labels
- reason:
- status: Status code as defined by grpc/codes in human-readable format, such as "OK" or "NotFound".
- offset_read: Whether the read request had a non-zero offset (e.g. Bazel retry resuming a partial read).
buildbuddy_proxy_capabilities_requests (Counter)
The number of requests served by a CapabilitiesServerProxy by gRPC status and cache hit/miss status.
Labels
- status: Status code as defined by grpc/codes. This is a numeric value; any non-zero code indicates an error.
- cache_status: Cache lookup result - One of: - "hit" - "miss" - "partial" (for batched RPCs where part of a request was cached) - Or "uncacheable" (for e.g. encrypted resources)
buildbuddy_proxy_capabilities_bytes (Counter)
The number of bytes served by a CapabilitiesServerProxy by gRPC status and cache hit/miss status.
Labels
- status: Status code as defined by grpc/codes. This is a numeric value; any non-zero code indicates an error.
- cache_status: Cache lookup result - One of: - "hit" - "miss" - "partial" (for batched RPCs where part of a request was cached) - Or "uncacheable" (for e.g. encrypted resources)
buildbuddy_proxy_content_addressable_storage_requests (Counter)
The number of requests served by a ContentAddressableStorageServerProxy by CAS operation and cache hit/miss status.
Labels
- op: ContentAddressableStorage Server operation: "FindMissingBlobs", "BatchUpdateBlobs", "BatchReadBlobs", or "GetTree".
- cache_status: Cache lookup result - One of: - "hit" - "miss" - "partial" (for batched RPCs where part of a request was cached) - Or "uncacheable" (for e.g. encrypted resources)
buildbuddy_proxy_content_addressable_storage_digests (Counter)
The number of digests served by a ContentAddressableStorageServerProxy by CAS operation and cache hit/miss status.
Labels
- op: ContentAddressableStorage Server operation: "FindMissingBlobs", "BatchUpdateBlobs", "BatchReadBlobs", or "GetTree".
- cache_status: Cache lookup result - One of: - "hit" - "miss" - "partial" (for batched RPCs where part of a request was cached) - Or "uncacheable" (for e.g. encrypted resources)
- compression: Describes the type of compression
buildbuddy_proxy_content_addressable_storage_bytes (Counter)
The number of bytes served by a ContentAddressableStorageServerProxy by CAS operation and cache hit/miss status.
Labels
- op: ContentAddressableStorage Server operation: "FindMissingBlobs", "BatchUpdateBlobs", "BatchReadBlobs", or "GetTree".
- cache_status: Cache lookup result - One of: - "hit" - "miss" - "partial" (for batched RPCs where part of a request was cached) - Or "uncacheable" (for e.g. encrypted resources)
- compression: Describes the type of compression
buildbuddy_proxy_find_missing_blobs_cache_lookups (Counter)
The number of digests looked up in the Cache Proxy's local FindMissingBlobs cache, by lookup status, whether the digests are CDC chunks, and result source.
'hit': served from the local cache. 'miss': the remote reported the digest present, so the lookup could have been a hit. 'uncacheable': the remote reported the digest missing; absence is never cached, so these lookups always require a remote check.
Labels
- cache_status: Cache lookup result - One of: - "hit" - "miss" - "partial" (for batched RPCs where part of a request was cached) - Or "uncacheable" (for e.g. encrypted resources)
- chunked: Whether the request was handled using chunking
- result_source: Source used by a cache proxy request to produce a response.
buildbuddy_proxy_remote_atime_updates (Counter)
The number of remote atime updates enqueued, with the outcome of the enqueue operation.
Labels
- group_id: Group (organization) ID associated with the request.
- status: There are a couple of places where we enqueue RPCs to be batched and sent asynchronously: the atime_updater and the hit_tracker_client. This label tracks the outcome of these enqueue operations. One of "enqueued", "duplicate", "dropped_too_many_updates", or "dropped_too_many_batches"
buildbuddy_proxy_remote_atime_update_requests (Counter)
The number of FindMissingBlobRequests sent to the remote cache to update remote blob atimes.
Labels
- group_id: Group (organization) ID associated with the request.
- status: Status code as defined by grpc/codes. This is a numeric value; any non-zero code indicates an error.
buildbuddy_proxy_remote_hit_tracker_updates (Counter)
The number of remote hit-tracker updates enqueued, with the outcome of the enqueue operation.
Labels
- group_id: Group (organization) ID associated with the request.
- status: There are a couple of places where we enqueue RPCs to be batched and sent asynchronously: the atime_updater and the hit_tracker_client. This label tracks the outcome of these enqueue operations. One of "enqueued", "duplicate", "dropped_too_many_updates", or "dropped_too_many_batches"
buildbuddy_proxy_remote_hit_tracker_requests (Histogram)
The number of HitTrackerService.Track RPCs sent to the remote hit-tracker service to record cache proxy cache hits.
Histogram values are the number of hit-updates per request.
Labels
- group_id: Group (organization) ID associated with the request.
- status: Status code as defined by grpc/codes. This is a numeric value; any non-zero code indicates an error.
buildbuddy_ociregistry_cache_download_size_bytes (Histogram)
Number of bytes downloaded from the cache by the OCI registry mirror
Labels
- oci_resource_type:
buildbuddy_ociregistry_cache_events (Counter)
Number of cache events handled.
Labels
- oci_resource_type:
- cache_event_type: Cache event type:
hit,miss, orupload.
buildbuddy_ociregistry_egress_size_bytes (Counter)
The number of bytes served, broken down by destination provider/region inferred from the peer IP.
Labels
- provider: Destination cloud provider inferred from the remote IP range:
aws,gcp, orother. - remote_region: Destination region inferred from the remote IP range, or
unknownif no known cloud range matches.
buildbuddy_ocifetcher_request_count (Counter)
Number of OCIFetcher requests by method, role, and status.
Labels
- method:
- role:
- status:
buildbuddy_ocifetcher_request_duration_usec (Histogram)
Duration of OCIFetcher requests by method and role, in microseconds.
Labels
- method:
- role:
buildbuddy_remote_execution_input_tree_setup_op_latency_usec (Histogram)
Latency of individual operations used for setting up input trees.
Labels
- op:
buildbuddy_disk_file_writer_in_progress_ops (Gauge)
Number of started, but not yet finished, FileWriter operations.
This number includes operations that are blocked on the concurrency limiter.
buildbuddy_disk_file_writer_tmp_file_bytes (Gauge)
Total size, in bytes, of temporary files currently staging writes for in-progress FileWriter operations.
Incremented as bytes are written to the temp file, and decremented when the temp file is committed to its final path or deleted.
Container image fetch metrics
buildbuddy_remote_execution_image_fetch_duration_usec (Histogram)
Duration of container image fetch attempts on executors, in microseconds.
Use the _count suffix for fetch counts.
Labels
- isolation: Effective workload isolation type used for an executed task, such as "docker", "podman", "firecracker", or "none".
- registry: Label name for the eTLD+1 of the container image registry.
- status: Status code as defined by grpc/codes. This is a numeric value; any non-zero code indicates an error.
- on_disk: Label name for whether the image was already on disk on the executor.
- has_creds: Label name for whether credentials were provided for the image fetch.
- trigger: Label name for what triggered the image fetch.
- use_oci_fetcher: Label name for whether the OCI fetcher service was used for the image fetch.
buildbuddy_grpc_client_rpcs_per_connection (Gauge)
A gauge measuring the number of pending RPCs per gRPC client connection, broken down by target, connection pool, and gRPC method.
Labels
- target:
- pool_id: Unique identifier for a gRPC client connection pool. For disambiguating between multiple pools connecting to the same target.
- grpc_method:
- connection_id: The index of a gRPC connection in a gRPC connection pool
buildbuddy_grpc_client_connection_count (Gauge)
Number of client gRPC connections (channelz sockets) observed per target, sampled from channelz.
Labels
- target:
buildbuddy_grpc_client_flow_control_blocked_connections (Gauge)
Number of client gRPC connections whose remote HTTP/2 flow-control (send) window is 0, per target.
A nonzero value means connections are send-blocked behind the peer's flow control, which head-of-line-blocks every stream on the connection.
Labels
- target:
buildbuddy_grpc_client_flow_control_window_bytes (Histogram)
Distribution of HTTP/2 flow-control window sizes across client gRPC connections, sampled periodically from channelz.
direction=remote is the window for data we send (0 means send-blocked); direction=local is the window for data we receive.
Labels
- target:
- direction: The direction of an HTTP/2 flow-control window relative to the local endpoint:
remote(the window for data we send) orlocal(the window for data we receive).
buildbuddy_grpc_client_connection_open_streams (Histogram)
Distribution of open HTTP/2 streams across client gRPC connections, sampled periodically from channelz.
Values near the peer's max-concurrent-streams limit (commonly 100) indicate connections at capacity.
Labels
- target:
buildbuddy_grpc_server_egress_bytes (Counter)
The number of gRPC server response bytes sent over the wire, broken down by gRPC method and destination provider/region inferred from the peer IP.
Note: this metric tracks gRPC payload bytes, which may be compressed, and does not include HTTP/2 framing or response headers.
Labels
- group_id: Group (organization) ID associated with the request.
- provider: Destination cloud provider inferred from the remote IP range:
aws,gcp, orother. - remote_region: Destination region inferred from the remote IP range, or
unknownif no known cloud range matches.
buildbuddy_grpc_server_ingress_bytes (Counter)
The number of gRPC server request bytes received over the wire, broken down by source provider/region inferred from the peer IP.
Note: this metric tracks gRPC payload bytes, which may be compressed, and does not include HTTP/2 framing or request headers.
Labels
- group_id: Group (organization) ID associated with the request.
- provider: Destination cloud provider inferred from the remote IP range:
aws,gcp, orother. - remote_region: Destination region inferred from the remote IP range, or
unknownif no known cloud range matches.
DNS server metrics
buildbuddy_dns_server_request_count (Counter)
The total number of DNS queries handled, by record type, response code, and recursive resolver provider.
Labels
- record_type: The DNS query (record) type, such as "A", "AAAA", "CNAME", or "MX". Unrecognized types are bucketed as "OTHER" to bound cardinality, since the type is client-controlled.
- rcode: The DNS response code, such as "NOERROR", "NXDOMAIN", or "FORMERR".
- resolver_provider: The provider operating the recursive resolver that sent the query, inferred from the transport peer's ASN. Values are bounded by the DNS server.
Examples
# DNS queries per second by record type
sum by (record_type) (rate(buildbuddy_dns_server_request_count[5m]))
# DNS queries per second by recursive resolver provider
sum by (resolver_provider) (rate(buildbuddy_dns_server_request_count[5m]))
# NXDOMAIN rate
sum(rate(buildbuddy_dns_server_request_count{rcode="NXDOMAIN"}[5m]))
/
sum(rate(buildbuddy_dns_server_request_count[5m]))
buildbuddy_dns_server_handler_duration_usec (Histogram)
Time to handle a DNS query, in microseconds, by record type.
Labels
- record_type: The DNS query (record) type, such as "A", "AAAA", "CNAME", or "MX". Unrecognized types are bucketed as "OTHER" to bound cardinality, since the type is client-controlled.
Examples
# Median DNS handler latency in the past 5 minutes
histogram_quantile(
0.5,
sum(rate(buildbuddy_dns_server_handler_duration_usec_bucket[5m])) by (le)
)
buildbuddy_dns_server_zone_serial (Gauge)
The SOA serial of each DNS zone currently served, by zone apex.
A zone that stops being served drops off the metric.
Labels
- dns_zone: The apex of a served DNS zone, such as "buildbuddy.io.". Zones come from operator-controlled zone files, so cardinality is bounded. Named "dns_zone" because ZoneLabel ("zone") is the availability zone of a node.
Examples
# Zones currently served, with their versions
max by (dns_zone) (buildbuddy_dns_server_zone_serial)
# Zones where replicas disagree on the served version
max by (dns_zone) (buildbuddy_dns_server_zone_serial)
!=
min by (dns_zone) (buildbuddy_dns_server_zone_serial)