-
Notifications
You must be signed in to change notification settings - Fork 182
Auto-configure prefix-cache-scorer parameters from engine metrics #1629
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Auto-configure prefix-cache-scorer parameters from engine metrics #1629
Conversation
✅ Deploy Preview for gateway-api-inference-extension ready!
To edit notification comments on pull requests, go to your Netlify project configuration. |
Hi @learner0810. Thanks for your PR. I'm waiting for a kubernetes-sigs member to verify that this patch is reasonable to test. If it is, they should reply with Once the patch is verified, the new status will be reflected by the I understand the commands that are listed here. Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes-sigs/prow repository. |
/assign @liu-cong |
@learner0810 do we have a similar metric in sglang? |
76971e1
to
d50488a
Compare
I looked up the sglang documentation and didn't find it, so it probably isn't there yet. |
cc @liu-cong PTAL |
LoraInfoWaitingAdaptersMetricName = "waiting_lora_adapters" | ||
LoraInfoMaxAdaptersMetricName = "max_lora" | ||
|
||
CacheConfigBlockSizeInfoMetricName = "block_size" |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
do you plan to add the cache size metric as well? maybe in a follow up PR? That's a much more useful metric than the block size. In fact, the block size doesn't necessarily need to be the same with vllm
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Yes, If IGW requires cache size metric, I'd be happy to add that in a subsequent pull request.
cc @elevran What's the status of the pluggable data layer? Can we use that to retrieve this new metric(s)? |
@liu-cong I was on PTO the past 2w. |
Excuse me, could you please advise me? What else should I change in my PR? |
The other approach for data layer extracting metrics should start from here. ref. But since it's not enabled, do you think we can merge this using the v1 approach first? @liu-cong @elevran And we can do the migration in another PR. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
/lgtm
The implementation adds to the metric code paths of the legacy and new data layer, which is fine until we fully migrate to the new data layer.
d50488a
to
93882e5
Compare
72e001e
to
20ec337
Compare
20ec337
to
dc8c13c
Compare
cc @liu-cong All tests are passing, PATL |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
/lgtm
/approve |
[APPROVALNOTIFIER] This PR is APPROVED This pull-request has been approved by: kfswain, learner0810 The full list of commands accepted by this bot can be found here. The pull request process is described here
Needs approval from an approver in each of these files:
Approvers can indicate their approval by writing |
What type of PR is this?
/kind feature
What this PR does / why we need it:
Auto-configure prefix-cache-scorer parameters from engine metrics
Which issue(s) this PR fixes:
Fixes # partial fix: #1512
Does this PR introduce a user-facing change?: