Getting a lock from a previous job fails sometimes

concourse / pool-resource

atomically manages the state of the world (e.g. external environments)

Apache License 2.0

56 stars 36 forks source link

Getting a lock from a previous job fails sometimes #5

Closed mhoran closed 9 years ago

mhoran commented 9 years ago

Similar to #2, we've seen the pool-resource fail to acquire a lock that has passed through a previous job. There seems to be a race condition where the checked-out lock is not "seen" by the pool resource by subsequent jobs. See https://diego.ci.cf-app.com/pipelines/main/jobs/ketchup-run-wats/builds/151 for an example. Interestingly, a subsequent run of the job passed with no changes.

concourse-bot commented 9 years ago

Hi there!

We use Pivotal Tracker to provide visibility into what our team is working on. A story for this issue has been automatically created.

The current status is as follows:

[ ] #101704536 Getting a lock from a previous job fails sometimes

This comment, as well as the labels on the issue, will be automatically updated as the status in Tracker changes.

mhoran commented 9 years ago

Looks like this was due to queueing up the same job twice, and the second time the job tried to run it failed to grab the lock that had already been checked in. Sorry for the noise!