[service] CI situation

[ericLemanissier]

@uilianries @SSE4 @jgsogo @danimtb
Is there a situation with the CI ? Several PR's keep failing unexpectedly since last Friday, for example:

• [gtk]: Enable gtk3 on windows and use shared glib when shared #11308
• quazip/xxx: Remove local _qt_major definition #11379
• glib 2.73.1 #11323
• [gdk-pixbuf] link against older glib only when building debug version #11357
• Qt: Add QMacStylePlugin under macOS #11321

I guess we should all stop to close/reopen PRs in order to trigger CI, because it seems to trigger itself? We are quite in the dark here, especially since the CI does not report "in progress" status, and the auto-trigger period seems to be unpredictable (several hours to several days).

[jwillikers]

I've also been unable to get boost/1.79.0: Patch Boost.SmartPtr to support a specific variant of PowerPC #11402 to pass in CI on Windows for several days, even though the file the patch applies to shouldn't even be built in CI since it is for PowerPC only.

[uilianries]

It's under investigation, the source is not clear yet, most errors come from Windows machine, but not specific error, like network or disk usage. Sorry for the delay.

[danimtb]

You are right, we have been experiencing some unexpected errors in some agents that could be caused by unresponsive windows machines when they are under a certain load. We have restarted the Jenkins instance behind CCI and some of the windows machines have been restarted as well.

Some of the PRs above are already running and the rest will follow them. Thanks a lot for notifying the issue and we will keep an eye on those jobs during the week.

[Croydon]

#11322 is also affected

[ericLemanissier]

also:

• fix pango requirements #11378
• harfbuzz 4.4.1 #11409

Does it help that we continue referencing the unexpected failures here, or is it an information you already have easily ?

EDIT: I guess https://github.com/conan-io/conan-center-index/pulls?q=is%3Aopen+is%3Apr+label%3A%22Unexpected+Error%22+-label%3Ainfrastructure+ gives a good enough overview

[uilianries]

@ericLemanissier we can filter them by the Github label Unexpected Error.

[jwillikers]

Looks like I'm seeing an issue for a missing pre-built binary for sqlite on a Windows build. It's not marked as an unexpected error: https://c3i.jfrog.io/c3i/misc/logs/pr/11395/11-configs/windows-visual_studio/nss/3.76.1//fc5e8d498438489803b4062553d0100c50f02f40-build.txt

[uilianries]

@jwillikers we are no longer building runtime MT + shared option.

[uilianries]

Back on the rails. Closing this one, the problem was related to Windows machines. Kudo to @danimtb

[theartful]

I still keep getting unexpected errors with different versions of gtk always on Windows, any idea why?
#11308

[uilianries]

@hesham-essam your case is more specific. Let's discuss in your PR. I gonna take a look.

[ericLemanissier]

@uilianries of all the PRs listed in this issue, a single one could be successfully built. There seem to be something else causing trouble again.

[uilianries]

@ericLemanissier I saw some linux machines were restarted. Last error we had were Windows machines ... Now we just need Mac to complete the trip of aces