Feature Request: Ability to clean up failed builds via the UI about bin HOT 3 CLOSED

When the filesystem used by the worker gets filled-up, we appear to get into an unrecoverable state with concourse itself throwing out errors that will make no sense to the uninitiated. e.g.:

• from a resource: hijack: Backend error: Exit status: 500, message: {"Type":"","Message":"mkdir /mnt/concourse/depot/8a0229e1-c899-44c0-5575-b01fbe314e9c/processes/d76dfcfc-548e-47b2-585c-dc1a2e54dac0: no space left on device","Handle":""}
• from a job: connection: process error: the exitcode file is empty
• from another job: mkdir /mnt/concourse/depot/37d776c6-4933-4187-5e4d-e95dfbc5ba99: no space left on device
• from another job: command not found

It seems like we should go further than allowing users to manually clean-up failed builds via the UI. It seems like concourse should be able to determine when the filesystem used by the worker is full and start removing artifacts from old failed builds - probably starting with the oldest.

This gives a better user experience by concourse cleaning up after itself above & beyond the normal cleanup that occurs after a successful build.

from bin.

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:

• #136961439 Feature Request: Ability to clean up failed builds via the UI

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

from bin.

Moved to concourse/concourse#1140

from bin.