I’ve set up a CI/CD pipeline with a runner using a docker image. The first stage of this pipeline has 2 jobs (unit-test and linter) that run in parallel. This pipeline has worked perfectly for 4 weeks.
Then one of the two jobs started failing randomly because of concurrency issues. When looking at the log, I see that the two jobs suddenly use the same concurrency number (always 0) for the runner:
Running on runner-nmsr1js--project-330-concurrent-0 via gitlab-srv
Previously, one of the jobs had concurrent-0 and the second one had concurrent-1
The runner config is quite simple, has not changed, and does not set any limit:
[[runners]] [runners.docker] image = "myimage" privileged=true disable_cache = false
I’ve tried to set a limit to the runner, global concurrency :
concurrent=6 [[runners]] limit = 4 request_concurrency = 4 [runners.docker] image = "myimage" privileged=true disable_cache = false
… without success
I am using a self-managed instance, version 14.6
Thanks for any suggestion that could help.