You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Affected Puppet, Ruby, OS and module versions/distributions
Any/all versions, platforms and combinations
How to reproduce (e.g Puppet code you use)
change a secret
change a config
run puppet
What are you seeing
gitlab-ctl reconfigure receives a refresh event from puppet twice, and therefore tries to execute twice. Eventually, these timeout because the the background chef processes from the first exec wind up still running and the second execution can't get the lock it needs for chef to to even begin. The Exec command eventually times out entirely, and the puppet run continues showing an error for the Exec['gitlab_reconfigure'], when in fact it may have succeeded.
Any additional information you'd like to impart
gitlab_reconfigure should only execute once per catalog run, and should execute after all the other resources in config.pp.
This doesn't happen often, and if we get rid of secrets (#213), there won't be anything that currently triggers this, but refactoring this so any and all "configs" are executed, before the gitlab_reconfigure runs, will allow us to more readily support new resources in the config manifest in the future (and I have a few in mind coming down the pike).
The text was updated successfully, but these errors were encountered:
LongLiveCHIEF
added a commit
to LongLiveCHIEF/puppet-gitlab
that referenced
this issue
Jun 2, 2018
Affected Puppet, Ruby, OS and module versions/distributions
Any/all versions, platforms and combinations
How to reproduce (e.g Puppet code you use)
What are you seeing
gitlab-ctl reconfigurereceives arefreshevent from puppet twice, and therefore tries to execute twice. Eventually, these timeout because the the background chef processes from the first exec wind up still running and the second execution can't get the lock it needs for chef to to even begin. The Exec command eventually times out entirely, and the puppet run continues showing an error for theExec['gitlab_reconfigure'], when in fact it may have succeeded.Any additional information you'd like to impart
gitlab_reconfigureshould only execute once per catalog run, and should execute after all the other resources inconfig.pp.This doesn't happen often, and if we get rid of secrets (#213), there won't be anything that currently triggers this, but refactoring this so any and all "configs" are executed, before the
gitlab_reconfigureruns, will allow us to more readily support new resources in the config manifest in the future (and I have a few in mind coming down the pike).The text was updated successfully, but these errors were encountered: