https://pantsbuild.org/ logo
Powered by
#general
  • a

    ancient-stone-50795

    10/16/2016, 1:44 PM
    Is it just my change or is something else conspiring to make shard 6 in travis take a long time to complete: https://travis-ci.org/pantsbuild/pants/builds/168026280 I've re-run it once.
  • a

    ancient-stone-50795

    10/16/2016, 1:45 PM
    I'm thinking its not just me, Benjy's PR had a similar issue https://travis-ci.org/pantsbuild/pants/builds/167994600
  • h

    happy-kitchen-89482

    10/16/2016, 4:09 PM
    Yeah, I’ve had the same problem
  • h

    happy-kitchen-89482

    10/16/2016, 4:09 PM
    time for a cache clean?
  • h

    happy-kitchen-89482

    10/16/2016, 4:54 PM
    I nuked the caches and restarted the timed-out shards
  • h

    happy-kitchen-89482

    10/16/2016, 4:54 PM
    on your build as well, eric
  • a

    ancient-stone-50795

    10/16/2016, 6:07 PM
    Thanks Benjy. Lets' see if that helps.
  • h

    happy-kitchen-89482

    10/16/2016, 6:11 PM
    It did not, sadly
  • w

    wide-energy-11069

    10/16/2016, 7:16 PM
    @ancient-stone-50795 the shard may be too big with the 50 min limit (https://github.com/travis-ci/travis-ci/issues/3810), would be a good idea to create one more shard?
  • a

    ancient-stone-50795

    10/16/2016, 7:33 PM
    @happy-kitchen-89482 It stinks that its grown by > 20 minutes for some unknown reason. Here's a build from yesterday: https://travis-ci.org/pantsbuild/pants/builds/167827929
  • h

    happy-kitchen-89482

    10/16/2016, 7:34 PM
    Why would that one shard suddenly leap to 50 minutes? It was the same as the others (~20-30) before
  • a

    ancient-stone-50795

    10/16/2016, 7:35 PM
    No, some of the other shards are longer too. Compare that one to mine: https://travis-ci.org/pantsbuild/pants/builds/168062004
  • a

    ancient-stone-50795

    10/16/2016, 7:35 PM
    I think we are just getting slower machines for some reasons? gotta run
  • h

    happy-kitchen-89482

    10/16/2016, 7:35 PM
    Copy code
    ============ slowest 3 test durations ============
                     503.02s call     tests/python/pants_test/backend/jvm/tasks/jvm_compile/zinc/test_zinc_compile_integration.py::ZincCompileIntegrationTest::test_zinc_fatal_warning
                     498.79s call     tests/python/pants_test/backend/jvm/tasks/jvm_compile/zinc/test_zinc_compile_integration.py::ZincCompileIntegrationWithZjars::test_zinc_fatal_warning
                     436.86s call     tests/python/pants_test/projects/test_testprojects_integration.py::TestProjectsIntegrationTest::test_testprojects_v2_engine
  • h

    happy-kitchen-89482

    10/16/2016, 7:36 PM
    Whereas the older builds had:
  • h

    happy-kitchen-89482

    10/16/2016, 7:36 PM
    Copy code
    ============ slowest 3 test durations ============
                     352.23s call     tests/python/pants_test/projects/test_testprojects_integration.py::TestProjectsIntegrationTest::test_testprojects_v2_engine
                     130.99s call     tests/python/pants_test/tasks/test_bootstrap_jvm_tools_integration.py::BootstrapJvmToolsIntegrationTest::test_zinc_tool_reuse_between_scala_and_java
                     71.55s call     tests/python/pants_test/backend/jvm/tasks/jvm_compile/zinc/test_zinc_compile_integration.py::ZincCompileIntegrationTest::test_zinc_fatal_warning
                     ========== 78 passed in 1360.90 seconds ==========
  • h

    happy-kitchen-89482

    10/16/2016, 7:37 PM
    Hmm, yeah, but everything got slower
  • h

    happy-kitchen-89482

    10/16/2016, 7:37 PM
    might be travis machines, as you say
  • b

    bored-art-40741

    10/16/2016, 8:26 PM
    “machines"
  • b

    bored-art-40741

    10/16/2016, 8:26 PM
    Given the times we were seeing for some process-local network timeout tests, they’re extremely saturated VMs
  • w

    wide-energy-11069

    10/16/2016, 10:27 PM
    Could be because it's free tier. Travis ci used the term 'fair use' https://travis-ci.com/plans
  • h

    happy-kitchen-89482

    10/16/2016, 10:54 PM
    Sure, but nothing changed on our end afaik
  • w

    wide-energy-11069

    10/16/2016, 11:09 PM
    if we added a lot of 
    @ensure_engine
    then those tests will be run twice essentially. although i don’t know the quantitive data on this.
  • a

    ancient-stone-50795

    10/16/2016, 11:51 PM
    but we didn't do that. The last build from master is fast
  • w

    wide-energy-11069

    10/17/2016, 1:25 AM
    reran shard 6 on master, got same error. the cache is likely playing a role. https://travis-ci.org/pantsbuild/pants/jobs/167827935
  • w

    wide-energy-11069

    10/17/2016, 1:26 AM
    was able to get it work with one additional shard https://travis-ci.org/wisechengyi/pants/builds/168099102
  • l

    limited-country-20626

    10/17/2016, 5:02 AM
    I had issues passing 
    exclude_target_regexp
    to 
    test-changed
    in 1.2.0rc1. https://github.com/pantsbuild/pants/issues/3977
  • a

    ancient-stone-50795

    10/17/2016, 10:58 AM
    Looks like killing the cache exposed some underlying issues:
  • a

    ancient-stone-50795

    10/17/2016, 10:59 AM
    https://travis-ci.org/pantsbuild/pants/jobs/168153820 These errors should have come out when we switched the Java platform default from 6 to 8 but we never saw them
  • c

    cuddly-garage-87447

    10/17/2016, 11:04 AM
    Hi, I'm working in a company building micro-services mostly written in java with the exception of 2 teams that are writing theirs in scala (between 10-20 services). We have multi repos in a typical setup: one git repo (in bitbucket) -> one jenkins pipeline (generated by jenkin's jobdsl). Would it make sense for the 2 scala teams (and only them) to share a common git mono-repo built with pants? The two teams occasionally share libraries. Would it make sense to go down to a mono-repo per team?
1...151415151516...2420Latest