concurrent docker-jobs failing for master (831d10)
When using the current master (1.1.0~beta.64.g831d102 (831d102)
) I'm seeing that for concurrent Docker-jobs, all but the last job is failing with an exit code 137
.
Steps to reproduce
-
ci-config
# .gitlab-ci.yml test1: tags: - docker script: - sleep 5 test2: tags: - docker script: - sleep 5
-
config.toml
concurrent = 5 [[runners]] name = "name" url = "https://example.org/ci" token = "token" executor = "docker" [runners.docker] tls_verify = true image = "ruby:2.1" privileged = false disable_cache = false volumes = ["/cache"]
-
build log
gitlab-ci-multi-runner 1.1.0~beta.64.g831d102 (831d102) Using Docker executor with image ruby:2.1 ... Running on runner-1489143c-project-26-concurrent-0 via vagrant-ubuntu-trusty-64... Fetching changes... HEAD is now at 2633126 Minimum From https://example.org/gert/ci-playground 2633126..d54b1ed issue/runner-exit -> origin/issue/runner-exit Checking out d54b1ed3 as issue/runner-exit... Previous HEAD position was 2633126... Minimum HEAD is now at d54b1ed... No image Checking cache for test1/issue/runner-exit/cache.zip... $ sleep 5 ERROR: Build failed with: exit code 137
GitLab version: 8.4.4
things tried
- Setting
concurrent
to 1 for runner: both jobs succeed. - Add a job
test3
with same ci-config: test1, and test2 fail. test3 succeeds. -
sleep
=>apt-get update
: fails as described above. -
sleep
=>ls
: all jobs succeed. - pick small image that would run bash (
tavisto/alpine-bash:latest
) and watch mem-usage: fails as described above. - ran on Darwin, and on Ubuntu-14.04 VM
- use runner
v1.0.4
: all jobs pass.
Running bisect singled out commit https://gitlab.com/gitlab-org/gitlab-ci-multi-runner/commit/003ad31df75dcf70a495d4544c43818fc99cd93f as the culprit.