Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Activity
    should remove any residual jobs from new PRs or webhooks in our orgs
    won't cover existing PRs
    Matthew R. Becker
    @beckermr
    seems like it is master to main
    even without a travis.yaml, travis goes back to the last valid commit and kicks off a build
    not sure if the above will help or not
    but let's try it
    so weird
    jakirkham
    @jakirkham
    Very strange. Would have thought those cases would be skipped. Anyways good to know why it was happening
    Tim D. Smith
    @tdsmith
    a librsvg build failed today after rustc hung for 5.5 hours on ppc64le -- any insights here? https://dev.azure.com/conda-forge/feedstock-builds/_build/results?buildId=487362&view=logs&j=8c139a40-380b-5518-bf6b-edcb18cbf319
    Isuru Fernando
    @isuruf
    @rpsene, some of the ppc64le jobs stop without any cause and the same job in aarch64 works just fine
    we are seeing this across multiple github repos
    Rafael Sene
    @rpsene
    hey @isuruf
    sorry for the delay, are you still facing issues with Travis?
    Isuru Fernando
    @isuruf
    @rpsene, no worries. Yes, it's still an issue
    Isuru Fernando
    @isuruf
    @rpsene, new issue on ppc64le travis-ci.
    Automatic restarts limited: Please try restarting this job later or contact support@travis-ci.com.
    doesn't happen on aarch64 travis-ci
    Leo Fang
    @leofang
    Looks like Travis is having issues? The aarch64 CI was working fine yesterday, but now it's not even kicked off:
    conda-forge/cusparselt-feedstock#18
    could it be conda-forge's credits ran out? lol
    Matthew R. Becker
    @beckermr
    we have credits
    I think this is their backend
    Leo Fang
    @leofang
    got it, so we'll need to wait...
    Leo Fang
    @leofang
    should we open an issue somewhere?
    Matthew R. Becker
    @beckermr
    no place to do that
    we can email travis help but asking @rpsene is usually more fruitful
    Leo Fang
    @leofang
    Matthew R. Becker
    @beckermr
    ohh yeah
    feel free to open an issue there
    Leo Fang
    @leofang
    Leo Fang
    @leofang
    it seems aarch64 CIs are back but not ppc64le
    Leo Fang
    @leofang
    looks like both are back
    jakirkham
    @jakirkham
    :tada: :tada: :tada:
    Leo Fang
    @leofang
    Now we're hitting Automatic restarts limited: Please try restarting this job later or contact support@travis-ci.com. again for aarch64...?
    Leo Fang
    @leofang
    looks like I can't restart job in Travis directly, must through GitHub
    Isuru Fernando
    @isuruf
    @rpsene, this is still an issue with ppc64le. ppc64le jobs take a long time to start (sometime 10 hours after they are scheduled) and sometimes get cancelled.
    jakirkham
    @jakirkham

    Ran into a recent job timeout here. Restarting doesn't seem to help

    https://app.travis-ci.com/github/conda-forge/ucx-split-feedstock/jobs/576077429

    ^ @rpsene would it be possible to raise this with Travis?
    Ben Mares
    @maresb

    On an autotick-bot PR, I'm getting an error with a completely empty log. Restarting didn't help:
    https://app.travis-ci.com/github/conda-forge/dasel-feedstock/jobs/576187993

    I have a ppc64le for the last version, and it only has 10 downloads, so I suspect that nobody would notice if a minor version is lagging by 1. Would it be okay to merge, despite the failing ppc64le build?

    3 replies
    jakirkham
    @jakirkham
    Should add I've seen a few other failures on feedstocks today. Tried to restart a few. It looks like a couple may be progressing, but a couple are still queued (so may run into the same issue)
    Ben Mares
    @maresb
    After I merged the aforementioned PR, all of the builds (incl. ppc) were successful :tada: ... so hopefully the problem is fixed, or maybe I got lucky.
    Francesco Biscani
    @bluescarni
    Hello! I was wondering if there's a plan for transitioning the PPC64 conda-forge builds to the new ABI in which long double is a true quadruple-precision type (rather than double-double)