-
Notifications
You must be signed in to change notification settings - Fork 26
Description
What happened:
KV cache test failed sometimes; ZQM tries to sub but the address is in use. It happens in CI.
See jobs:
https://github.com/llm-d/llm-d-inference-sim/actions/runs/17185865922/job/48754776272#step:8:131
https://github.com/llm-d/llm-d-inference-sim/actions/runs/17186323665/job/48755883796
Line:
subEndpoint = "tcp://*:5557" |
Guess:
The executor that runs the job has side process occupied the addr *:5557
The addr is not "fully" released
Proposal:
@pancak3 - use 127.0.0.1
in tests for both sub and pub
@irar2 - look for available port
What you expected to happen:
Pass, successfully sub the addr
How to reproduce it (as minimally and precisely as possible):
Run test, it occasionally fails
Anything else we need to know?:
N/A
Environment:
N/A