Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Activity
    Leo Fang
    @leofang
    no idea why it's terminated without any error
    this one has the same symptom as that from yesterday
    https://travis-ci.com/github/conda-forge/openmpi-feedstock/builds/220971585
    Leo Fang
    @leofang
    jayfurmanek
    @jayfurmanek
    I think that Travis container is either running out of memory, or is hitting low ulimits for number of processes. Maybe try changing the -j in the recipe to a low static number to see if that helps:
    https://github.com/conda-forge/openmpi-feedstock/blob/master/recipe/build-mpi.sh#L154
    setting -j to the numer of "cpus" is not always appropriate on ppc64le since there can be up to 8 threads per core on some machines - and those show up as "cpus"
    I doubt Travis lets you set any ulimits in jobs
    Leo Fang
    @leofang
    nah, the build would run too long then, not worth it
    jayfurmanek
    @jayfurmanek
    but there may be too many threads kicked off on the ppc machine - it may actually be slower with too many.
    Leo Fang
    @leofang
    It used to work just fine, given the recent status of travis I incline to think it's not the recipe's problem
    notice that the 2nd time it failed after 1min --- we're not even spinning that many threads yet!
    it literally just received the job
    jayfurmanek
    @jayfurmanek
    yeah- the travis ppc64le nodes dont have much resources it seems.Maybe there are other users at the moment affecting things. Hard to say. My suggestion above was to maybe work around the travis limits, not a highlight of a general recipe problem
    Isuru Fernando
    @isuruf
    we are setting CPU_COUNT=2 on travis-ci
    jayfurmanek
    @jayfurmanek
    ah
    Leo Fang
    @leofang
    yeah I think the same is done on Azure
    now that Isuru mentioned
    jakirkham
    @jakirkham

    Are we sure it is related to CPU_COUNT? This fails well before that is relevant

    https://travis-ci.com/github/conda-forge/openmpi-feedstock/builds/220978820

    Isuru Fernando
    @isuruf
    @jakirkham, no it's not. Please read the whole thread
    Leo Fang
    @leofang
    It's ok, thanks John for working with me. Let me mention that interestingly all merge commits passes Travis in one shot, no retry needed. I think Travis is probably just unhappy testing out my PRs 😂
    Wolf Vollprecht
    @wolfv
    jakirkham
    @jakirkham
    @rpsene we are seeing builds getting terminated again. Here's a recent link ( https://travis-ci.com/github/conda-forge/matplotlib-feedstock/jobs/495320303 ). Though there are several
    jakirkham
    @jakirkham
    @rpsene ?
    Rafael Sene
    @rpsene
    oops, missed this one.
    let me check
    unning conda build
    travis_fold:start:Runningcondabuild\r+ conda build /home/conda/recipe_root -m /home/conda/feedstock_root/.ci_support/linux_ppc64le_numpy1.17python3.6.____cpython.yaml --suppress-variables --clobber-file /home/conda/feedstock_root/.ci_support/clobber_linux_ppc64le_numpy1.17python3.6.____cpython.yaml
    No numpy version specified in conda_build_config.yaml.  Falling back to default numpy value of 1.16
    WARNING:conda_build.metadata:No numpy version specified in conda_build_config.yaml.  Falling back to default numpy value of 1.16
    Adding in variants from internal_defaults
    INFO:conda_build.variants:Adding in variants from internal_defaults
    Adding in variants from /home/conda/feedstock_root/.ci_support/linux_ppc64le_numpy1.17python3.6.____cpython.yaml
    INFO:conda_build.variants:Adding in variants from /home/conda/feedstock_root/.ci_support/linux_ppc64le_numpy1.17python3.6.____cpython.yaml
    Attempting to finalize metadata for matplotlib-base
    INFO:conda_build.metadata:Attempting to finalize metadata for matplotlib-base
    Collecting package metadata (repodata.json): ...working...
    what is expected here ?
    jakirkham
    @jakirkham
    That one may have gotten restarted. Here's a more recent one ( https://travis-ci.com/github/conda-forge/opencv-feedstock/jobs/495510514 )
    All good. Thanks again for your help here :slight_smile:
    Rafael Sene
    @rpsene
    I didn’t touch anything this time.
    nor poked anyone
    let’s take as a miracle for now :D
    jakirkham
    @jakirkham
    Sorry not following
    Rafael Sene
    @rpsene
    All good. Thanks again for your help here :slight_smile:
    jakirkham
    @jakirkham
    It seems like jobs are dying at various stages, but without any indication as to why
    Yeah actually looking at the job above it still demonstrates the issue
    Does that make sense @rpsene ?
    Rafael Sene
    @rpsene
    let me poke travis, not sure what is going on
    jakirkham
    @jakirkham
    Ok thanks :)
    Wolf Vollprecht
    @wolfv
    @rpsene travis is still failing randomly ... this situation hasn't really improved in the past 3 months as far as I can tell
    ok I have to take it back a little, seems like most of the recent builds are fine:
    Screenshot 2021-04-03 at 10.42.09.png
    and octomap builds fine after merging ... still some random failures here and there it seems
    Rafael Sene
    @rpsene
    I will report this
    Isuru Fernando
    @isuruf
    no space left on device. error is back
    Uwe L. Korn
    @xhochy
    Seeing more no space left on device in the last days: https://travis-ci.com/github/conda-forge/r-lightgbm-feedstock/jobs/498078864
    Rafael Sene
    @rpsene
    poking Travis folks now
    Rafael Sene
    @rpsene
    @xhochy @isuruf Travis reported a fix was applied
    Isuru Fernando
    @isuruf
    Thanks