-
Notifications
You must be signed in to change notification settings - Fork 751
docs: document LMCache Prometheus metrics harmless error for vLLM 0.12.0 #5026
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
docs: document LMCache Prometheus metrics harmless error for vLLM 0.12.0 #5026
Conversation
Signed-off-by: Keiven Chang <[email protected]>
WalkthroughTwo vLLM documentation files were updated with troubleshooting sections. LMCache Integration guide now details PrometheusLogger and PROMETHEUS_MULTIPROC_DIR issues with mitigations. Prometheus metrics documentation references the new troubleshooting guidance. No functional code changes introduced. Changes
Estimated code review effort🎯 1 (Trivial) | ⏱️ ~5 minutes
Poem
Pre-merge checks✅ Passed checks (3 passed)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
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.
Actionable comments posted: 1
📜 Review details
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (2)
docs/backends/vllm/LMCache_Integration.md(1 hunks)docs/backends/vllm/prometheus.md(1 hunks)
🧰 Additional context used
🧠 Learnings (1)
📓 Common learnings
Learnt from: keivenchang
Repo: ai-dynamo/dynamo PR: 3035
File: lib/runtime/src/metrics/prometheus_names.rs:49-53
Timestamp: 2025-09-16T00:26:37.092Z
Learning: keivenchang prefers consistency in metric naming standardization over strict adherence to Prometheus conventions about gauge vs counter suffixes. When standardizing metrics naming, prioritize consistency across the codebase rather than technical pedantry about individual metric type conventions.
Learnt from: keivenchang
Repo: ai-dynamo/dynamo PR: 3051
File: container/templates/Dockerfile.trtllm.j2:424-437
Timestamp: 2025-09-16T17:16:03.785Z
Learning: keivenchang prioritizes maintaining exact backward compatibility during migration/refactoring PRs, even when bugs are identified in the original code. Fixes should be deferred to separate PRs after the migration is complete.
🪛 markdownlint-cli2 (0.18.1)
docs/backends/vllm/LMCache_Integration.md
169-169: Fenced code blocks should have a language specified
(MD040, fenced-code-language)
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (1)
- GitHub Check: Build and Test - dynamo
🔇 Additional comments (3)
docs/backends/vllm/prometheus.md (1)
132-137: Excellent cross-referencing and section structure.The new Troubleshooting subsection appropriately directs users to the detailed LMCache Integration Guide while maintaining a concise overview in the Prometheus metrics documentation. The placement and content are well-organized.
docs/backends/vllm/LMCache_Integration.md (2)
163-193: Excellent documentation of the PrometheusLogger singleton issue.The PrometheusLogger section is well-structured and provides essential context: version-specific behavior (v0.12.0), clear cause explanation, impact assessment (log-only, non-blocking), concrete reproduction steps, and practical mitigation. The note distinguishing this issue's vLLM version scope is particularly helpful.
194-200: Clear guidance for PROMETHEUS_MULTIPROC_DIR in Dynamo context.The PROMETHEUS_MULTIPROC_DIR section appropriately distinguishes between user-managed and Dynamo-managed scenarios, with clear remediation steps. Directing users to check their shell/container environment is practical and actionable.
Signed-off-by: Keiven Chang <[email protected]>
f45d3e9 to
001ef15
Compare
Overview:
Document LMCache + vLLM (0.12.0) metrics/log troubleshooting.
Details:
docs/backends/vllm/LMCache_Integration.md:PrometheusLogger ... different metadata(log-only; 0.12.0 repro, not seen on 0.11.0)PROMETHEUS_MULTIPROC_DIRwarning guidanceWhere should the reviewer start?
docs/backends/vllm/LMCache_Integration.md
Related Issues
DIS-1172
/coderabbit profile chill
Summary by CodeRabbit
✏️ Tip: You can customize this high-level summary in your review settings.