Fix flakey OOP log test

[adam-cattermole]

Exposed waiting on the waitgroup instead of the fixed time which seems to be inconsistent/insufficient in the runners

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 75.14%. Comparing base (41a4f2b) to head (ccc7e0b).
⚠️ Report is 11 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1744      +/-   ##
==========================================
+ Coverage   74.91%   75.14%   +0.22%     
==========================================
  Files         120      120              
  Lines       10287    10369      +82     
==========================================
+ Hits         7707     7792      +85     
- Misses       2206     2211       +5     
+ Partials      374      366       -8     

Flag	Coverage Δ
bare-k8s-integration	24.27% <0.00%> (+0.58%)	⬆️
controllers-integration	59.35% <0.00%> (-0.10%)	⬇️
envoygateway-integration	44.37% <0.00%> (+0.01%)	⬆️
gatewayapi-integration	19.21% <0.00%> (+0.25%)	⬆️
istio-integration	48.97% <0.00%> (+0.20%)	⬆️
unit	22.17% <100.00%> (-0.16%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

Components	Coverage Δ
api (u)	85.88% <ø> (ø)
internal (u)	76.91% <89.81%> (+0.24%)	⬆️
pkg (u)	31.00% <ø> (ø)

Files with missing lines	Coverage Δ
internal/extension/oop.go	49.12% <100.00%> (+0.90%)	⬆️

... and 9 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[didierofrivia]

much better approach! 🥇

[adam-cattermole]

Damn it isn't fixed; managed to get it to fail on the 9th run..

[adam-cattermole]

I think this is the problem.. I managed to reproduce:

"msg"="failed to read stderr" "error"="read |0: file already closed"

I think the cmd.Wait is closing the process and we're losing the stderr before the scanner gets scheduled/reads it

[adam-cattermole]

Okay, if you run this you should be able to get it to reproduce the error (we use separate processes so that it makes the machine more busy):

cd internal/extension && for i in {1..16}; do go test -run TestOOPExtensionForwardsLog -count=1000 -parallel=16 -failfast & done; wait

I seem to be able to pretty reliably get this to error on main but haven't hit it yet with the latest changes

(They're quite a bit slower to run on main because that 50ms wait is adding up across all runs)

[KevFan]

Nice! Ran the command and the CI a good few times and all looking good 💯