Closed Bug 1917711 Opened 1 year ago Closed 1 year ago

macOS webrender ("WR") tasks fail on almost all machines: abort: cannot enable sparse profile on existing non-sparse checkout

Categories

(Developer Services :: Mercurial: robustcheckout, defect)

defect

Tracking

(Not tracked)

RESOLVED FIXED

People

(Reporter: aryx, Unassigned)

References

()

Details

(Keywords: intermittent-failure)

See https://treeherder.mozilla.org/jobs?repo=autoland&revision=71eb0e99bd8debcc45397f7b5c5510e7ed90cf12&selectedTaskRun=ER30dEyOSsS1Du0kkKxIDQ.0 which had one WR(cargotest) run which succeed and many failing ones after 9 days without a run (task only scheduled if files in gfx/wr/ change).

The WR tasks use a sparse checkout profile but also try to use the existing shared checkout and bail because it is not a sparse one. It's unknown why almost every macOS hardware machine is affected now.

Connor, could you investigate?

When i've seen these errors in the past it was because the task that was supposed to do the initial clone and sparse config failed at some stage after creating the repo but before setting the sparse config, and left the cache corrupted.

This is also happening frequently with the macOS startup tests, which blocks release graphs.

:pmoore, do you have any idea what's going on here? Is there a cache directory that's shared between tasks?

Do the task_dirs in the cache directory ever get cleaned up?

Flags: needinfo?(pmoore)

I'm working with Ryan on open macOS issues now, will report back.

Flags: needinfo?(rcurran)
Flags: needinfo?(pmoore)
Flags: needinfo?(aryx.bugmail)

:aryx do we feel good about closing this?

Flags: needinfo?(rcurran)

Okay to close it. Thanks for working on this.

Status: NEW → RESOLVED
Closed: 1 year ago
Flags: needinfo?(aryx.bugmail)
Resolution: --- → FIXED
You need to log in before you can comment on or make changes to this bug.