chore(engine): add base set of new scheduler metrics #19913
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Observing the scheduler will require accessing workerConn concurrently from the /metrics endpoint and the connection handler. To prepare for this, all state handling for workerConn has been moved into methods, and all fields are now protected by a mutex. Signed-off-by: Robert Fratto <[email protected]>
Add a base set of scheduler metrics for monitoring: * loki_scheduler_tasks_total: Counter for task transitions (by state) * loki_scheduler_tasks_inflight: Gauge for current tasks (by state) * loki_scheduler_streams_total: Counter for stream transitions (by state) * loki_scheduler_streams_inflight: Gaguge for current streams (by state) * loki_scheudler_workers: Gauge for current workers (by state) * loki_scheduler_threads: Gauge for current threads (by state) * loki_scheduler_connections_total: Counter for total incoming conns * loki_scheduler_connections_active: Gauge for current active conns * loki_scheduler_task_queue_seconds: Tracks how long tasks are enqueued * loki_scheduler_task_exec_seconds: Tracks how long tasks take to finish Worker and thread state is a new concept, introduced for monitoring. They are in one of the following states: * busy: A task is executing * ready: Tasks have been requested * idle: No tasks have been requested or are running The overall worker state is inferred from the state of its threads, in precedence order of busy > ready > idle. Signed-off-by: Robert Fratto <[email protected]>
ashwanthgoli
approved these changes
Nov 19, 2025
| sched: sched, | ||
|
|
||
| tasksInflight: prometheus.NewDesc( | ||
| "loki_scheduler_tasks_inflight", |
Contributor
There was a problem hiding this comment.
nit: can be confused with the old scheduler, maybe loki_engine_scheduler_*
Changes the prefix as suggested to avoid confusion with the loki_query_scheduler_ prefix. Signed-off-by: Robert Fratto <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Note
I had to refactor
workerConna little for this; I isolated those changes into the first commit for easier review.This PR adds a base set of scheduler metrics for monitoring:
loki_engine_scheduler_tasks_total: Counter for task transitions (by state)loki_engine_scheduler_tasks_inflight: Gauge for current tasks (by state)loki_engine_scheduler_streams_total: Counter for stream transitions (by state)loki_engine_scheduler_streams_inflight: Gaguge for current streams (by state)loki_engine_scheduler_workers: Gauge for current workers (by state)loki_engine_scheduler_threads: Gauge for current threads (by state)loki_engine_scheduler_connections_total: Counter for total incoming connsloki_engine_scheduler_connections_active: Gauge for current active connsloki_engine_scheduler_task_queue_seconds: Tracks how long tasks are enqueuedloki_engine_scheduler_task_exec_seconds: Tracks how long tasks take to finishWorker and thread state is a new concept, introduced for monitoring. They are in one of the following states:
busy: A task is executingready: Tasks have been requestedidle: No tasks have been requested or are runningThe overall worker state is inferred from the state of its threads, in precedence order of
busy>ready>idle.