Logs of failing deployments cannot be retrieved anymore with newest cf

[stefanlay]

With loggregator-release v107.0.0 the RecentLogsHandler was removed from the trafficcontroller. It was also removed in the noaa client.

This endpoint is still used in the appdeployers and the v3appdeployers code to retrieve logs when application staging failed.

Instead of the noaa client and the trafficcontroller recentlogs endpoint the LogCache API endpoint GET /api/v1/read/<source-id>?envelope_types=LOG&limit=? should be used.

[chombium]

There is a go-log-cache client which can be used instead of the NOAA Client ;)

[chombium]

Hi @loafoe, @Thanhphan1147,
we (me and @ZPascal) will start working on this issue. Can you please assign us?

TIA,
Jovan

[loafoe]

@chombium any progress?

[chombium]

Hi @loafoe,
I've been really busy lately :( I'll give my best to open a PR in the next week or two.

[ZPascal]

Hi @loafoe, FYI we've got currently the problem that the requirements of the log-cache functionality are not available inside the replaced community release of the CF CLI implementation. Could you please describe the circumstances for the replacement process? cc: @chombium

[chombium]

I've checked the state of the state of the cloudfoundry-cli in the cloudfoundry-community org and the state is 2667 commits behind cloudfoundry/cli:main. We could try exchanging the cli library with the latest upstream, but it would be hard to tell what other things might go wrong and if this project has enough test.
@loafoe We can try to use the upstream cloudfoundry-cli from code.cloudfoundry, practically without the replace in go.mod, but we don't know if we're going to break something...