coulomb/resource-control
5
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Review capped Vergabe comparison without changing allocation

Browse source 
Assistant: codex
Assistant-Model: gpt-6.1-sol
Assistant-Session: 01a15374-f59f-7a31-86d2-d23b5d2ed181
This commit is contained in:
tegwick 2026-10-19 15:20:33 +02:00
parent c6dcad5a58
commit e3e996fc95
2 changed files with 21 additions and 2 deletions
Download patch file Download diff file Expand all files Collapse all files

17
docs/evidence/2026-10-19-custodian-allocation-review.md
View file

@ -39,3 +39,20 @@ owners for mas/playwright/rapp-qonto; representative shared-service and build-ru
demand; host/rollout margin; current contention and throttle evidence. Historical
portfolio utilization remains dated2026-08-11 and is not silently refreshed by this
reservation snapshot. RESOURCE-WP-0008 keeps the allocation review active.
## Capped comparison received
Vergabe commit `bfd1a2d`, `container-comparison.json` and retained raw samples
complete the first isolated 1000m/1Gi comparison: 1552 ordinary operations and
516 transfers, zero errors, ordinary p95 412ms and transfer p95 158ms. Coverage
is complete, OOM false. Mean app CPU was 940m, with 1449/1807 periods throttled;
peak cgroup memory was 426MiB including page cache. Separate database CPU p95
was 60.63% of one core. This supports retaining the existing limits provisionally
for the tested mix. It supplies no reason to raise the CPU limit or reduce the
memory limit; the earlier unconstrained envelope remains rejected.
The 60m reservation remains unvalidated under shared contention. The owner is
extending the same configurable workload with edits, startup/restart and burst
phases. Target-release/hardware and shared-service observations remain missing.
RESOURCE-WP-0008-T02 stays in progress; no allocation or spending changed.

6
workplans/RESOURCE-WP-0008-custodian-allocation-review.md
View file

@ -47,8 +47,10 @@ priority: high
state_hub_task_id: "1f538441-8b45-5d37-aeb3-345204b75117"
```
Consume the in-progress isolated app-container comparison at the existing
1000m CPU and 1Gi memory limits. The 60m request is a contention share/reservation,
Initial comparison received at Vergabe `bfd1a2d`: zero errors and ordinary p95
412ms under the existing 1000m CPU and 1Gi memory limits. The retained review
supports keeping those limits provisionally; edits, startup/restart, bursts and
target-runtime contention still need coverage. Consume those remaining results. The 60m request is a contention share/reservation,
not a CPU cap. Compare complete latency/error and CPU/throttle/memory evidence,
retain synthetic pacing assumptions and configurable safety factors, and state
what the sample cannot establish about production contention and shared services.