server: monitoring - add /metrics prometheus compatible endpoint #5708
Conversation
This file contains 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
|
@ggerganov I understand we want to globally refactor the server code, although this is a small change with no logic added in order to help people monitoring the server performances. Also I would be happy to submit an end-to-end kubernetes example later on based on this. |
ggerganov
approved these changes
Feb 25, 2024
ngxson
reviewed
Feb 25, 2024
…one call thread is notified
phymbert
commented
Feb 25, 2024
| @@ -441,7 +441,7 @@ struct llama_server_response { | |||
| { | |||
| LOG_VERBOSE("queue_results.push_back", {}); | |||
| queue_results.push_back(result); | |||
| condition_results.notify_one(); | |||
| condition_results.notify_all(); | |||
There was a problem hiding this comment.
@ggerganov @ngxson I was looking for a while why health/metrics/slots endpoints were blocking when a completion request is processing: only one caller thread was notified. Please confirm my fix.
There was a problem hiding this comment.
Thanks for noticing, that's LGTM. Seems like this is the root cause of many stability issues of concurrent requests.
jordankanter
pushed a commit
to jordankanter/llama.cpp
that referenced
this pull request
Mar 13, 2024
…ible endpoint (ggml-org#5708) * server: monitoring - add /metrics prometheus compatible endpoint * server: concurrency issue, when 2 task are waiting for results, only one call thread is notified * server: metrics - move to a dedicated struct
hodlen
pushed a commit
to hodlen/llama.cpp
that referenced
this pull request
Apr 1, 2024
…ible endpoint (ggml-org#5708) * server: monitoring - add /metrics prometheus compatible endpoint * server: concurrency issue, when 2 task are waiting for results, only one call thread is notified * server: metrics - move to a dedicated struct
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
Motivation
Monitoring inference performance in time series dashboard is a must have for LLM.
Prometheus format is the standard implementation to expose metrics.
Other LLM inference server offer this feature, for example vLLM.
Changes
/metrics--metricsoption