Skip to content

fix(metrics): expose scheduler and resource budgets - #315

Merged
ketor merged 1 commit into
dingodb:mainfrom
ketor:feat/hicache-client-metrics-v214
Aug 11, 2026
Merged

fix(metrics): expose scheduler and resource budgets#315
ketor merged 1 commit into
dingodb:mainfrom
ketor:feat/hicache-client-metrics-v214

Conversation

@ketor

@ketor ketor commented Aug 11, 2026

Copy link
Copy Markdown
Contributor

Problem

The native v2.14 client snapshot exports read-scheduler and process-wide RDMA resource metrics, but CLIENT_METRIC_SPECS silently drops them. SGLang/HiCache /metrics therefore cannot show queue delay, pending/active shards, budget saturation, or admission timeouts.

Fix

  • allow-list all 5 read-scheduler families
  • allow-list all 5 resource-budget families with bounded kind=used|limit labels
  • include the adjacent endpoint-cache hit/miss/eviction counters
  • add parser, label, counter-delta, gauge-update, and unknown-family regression coverage

Verification

  • focused HiCache poller tests: 7 PASS
  • native snapshot delta test: PASS
  • release contract: 3 PASS
  • complete local CTest: 405 PASS, 284 conditional SKIP, 0 FAIL
  • exhaustive client snapshot comparison: no remaining non-histogram client metric omissions

@ketor
ketor added this pull request to the merge queue Aug 11, 2026
Merged via the queue into dingodb:main with commit 9e9f97f Aug 11, 2026
8 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant