Opened at 2009-06-21T20:18:08Z
Closed at 2010-06-13T16:11:29Z
#744 closed defect (cannot reproduce)
leasecrawler test failure
| Reported by: | warner | Owned by: | warner |
|---|---|---|---|
| Priority: | minor | Milestone: | eventually |
| Component: | code | Version: | 1.4.1 |
| Keywords: | test | Cc: | |
| Launchpad Bug: |
Description
Zooko noticed an (intermittent?) failure on our cygwin buildslave in the lease-crawler tests:
http://allmydata.org/buildbot/builders/cygwin/builds/1968/steps/test/logs/stdio
It suggests that test_storage.LeaseCrawler.test_expire_cutoff_date has a synchronization step that's being skipped or missed somehow, such that an assertion was checked before the crawler had gotten far enough to fulfill it. I'd thought that I'd make this test pretty immune to timing/load-variation problems, but maybe I missed something.
Change History (3)
comment:1 Changed at 2009-06-21T20:18:18Z by warner
- Owner changed from somebody to warner
- Status changed from new to assigned
comment:2 Changed at 2010-06-12T22:14:39Z by davidsarah
- Keywords test added
comment:3 Changed at 2010-06-13T16:11:29Z by zooko
- Resolution set to cannot reproduce
- Status changed from assigned to closed
I don't think we had flogging turned on in those cygwin unit tests. Since the cygwin buildslave was turned off (I think it is now taking up space in Brian's closet) this is not currently reproducible. If someone runs the unit tests on cygwin and finds this problem then they can re-open this ticket.
Duplicate of #923?