server: metrics: fix bucket reset on /health, add few more metrics#5937
Merged
server: metrics: fix bucket reset on /health, add few more metrics#5937
Conversation
…s_predicted_seconds_total, reset bucket only on /metrics. Fix values cast to int. Add Process-Start-Time-Unix header. Closes #5850
ggerganov
approved these changes
Mar 8, 2024
Contributor
|
The server executable won't even launch for me from the merge commit of this PR onwards... (windows 10, MSVC build, any backend) |
Collaborator
Author
|
I doubt there is anything related. This PR is about metrics only. or maybe |
Contributor
|
I will try again tomorrow, but the program dies before printing anything to the standard output. No error message either. But I'm pretty sure these changes somehow caused the issue, because reverting to commit e457fb3 fixes it for me. |
Contributor
|
It got fixed by #5964, thank you @ggerganov |
hazelnutcloud
pushed a commit
to hazelnutcloud/llama.cpp
that referenced
this pull request
Mar 10, 2024
…s_predicted_seconds_total, reset bucket only on /metrics. Fix values cast to int. Add Process-Start-Time-Unix header. (ggml-org#5937) Closes ggml-org#5850
NeoZhangJianyu
pushed a commit
to NeoZhangJianyu/llama.cpp
that referenced
this pull request
Mar 12, 2024
…s_predicted_seconds_total, reset bucket only on /metrics. Fix values cast to int. Add Process-Start-Time-Unix header. (ggml-org#5937) Closes ggml-org#5850
jordankanter
pushed a commit
to jordankanter/llama.cpp
that referenced
this pull request
Mar 13, 2024
…s_predicted_seconds_total, reset bucket only on /metrics. Fix values cast to int. Add Process-Start-Time-Unix header. (ggml-org#5937) Closes ggml-org#5850
|
@phymbert Shouldn't the added metrics be document at https://github.com/ggerganov/llama.cpp/blob/968967376dc2c018d29f897c4883d335bbf384fb/examples/server/README.md?plain=1#L527-L533? |
Seunghhon
pushed a commit
to Seunghhon/llama.cpp
that referenced
this pull request
Apr 26, 2026
…s_predicted_seconds_total, reset bucket only on /metrics. Fix values cast to int. Add Process-Start-Time-Unix header. (ggml-org#5937) Closes ggml-org#5850
phuongncn
pushed a commit
to phuongncn/llama.cpp-gx10-dgx-sparks-deepseekv4
that referenced
this pull request
Apr 28, 2026
…s_predicted_seconds_total, reset bucket only on /metrics. Fix values cast to int. Add Process-Start-Time-Unix header. (ggml-org#5937) Closes ggml-org#5850
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 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.
Context
Server
/metricsendpoint share the same task event as/health:TASK_TYPE_METRICS. It means metrics are reset on both calls.The
Process-Start-Time-Unixhttp response header is not set.Add total prompt and generation time in order to get the global server throughput for all slots.
Closes #5850