Increase WOODPECKER_FORGE_TIMEOUT to fix config fetching for GitLab

[pat-s]

fix #4260

Config fetching on GL takes a bit longer due to the subgroup fetching and permissions checks.
In my case, I had 1/6 repos failing with the default of 3s. The repo only had two workflow files. Both were somewhat complex.

I assume that the workflow parsing combined with the subgroup checks exceeded the 3s here.
After bumping WOODPECKER_FORGE_TIMEOUT to 5s, it worked.
Getting the initial config fetched seems to speed up future config pull as afterwards I was able to reduce WOODPECKER_FORGE_TIMEOUT again to 3s.

However, to avoid such issues for users I think the default value should be bumped to (at least) 5s.

Additionally, I improved some of the log/trace outputs which helped me a bit along the way.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 26.52%. Comparing base (ddd55ee) to head (dfc3420).
Report is 26 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #4262      +/-   ##
==========================================
+ Coverage   26.47%   26.52%   +0.05%     
==========================================
  Files         376      377       +1     
  Lines       27465    27451      -14     
==========================================
+ Hits         7270     7282      +12     
+ Misses      19530    19504      -26     
  Partials      665      665              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[woodpecker-bot]

Deployment of preview was successful: https://woodpecker-ci-woodpecker-pr-4262.surge.sh

[pat-s]

Hmm while the build now starts, the webhook execution in this repo still fails with

and GL disables the webhook after a few executions which resulted in a timeout.

[6543]

still we can increase it a bit more ...

[pat-s]

I don't think the timeout is the issue here. Even with 20s the GL webhook quits after ~ 10s.
I assume something is broken with the response?

The build is starting nevertheless, so only the response to the webhook is somehow not being sent back in time.