Note: This is a public test instance of Red Hat Bugzilla. The data contained within is a snapshot of the live data so any changes you make will not be reflected in the production Bugzilla. Email is disabled so feel free to test any aspect of the site that you want. File any problems you find or give feedback at bugzilla.redhat.com.
Bug 1141451 - Failures in tests: /dev/sda1: No such file or directory
Summary: Failures in tests: /dev/sda1: No such file or directory
Keywords:
Status: CLOSED UPSTREAM
Alias: None
Product: Virtualization Tools
Classification: Community
Component: libguestfs
Version: unspecified
Hardware: Unspecified
OS: Unspecified
unspecified
unspecified
Target Milestone: ---
Assignee: Richard W.M. Jones
QA Contact:
URL:
Whiteboard:
Depends On:
Blocks: TRACKER-bugs-affecting-libguestfs
TreeView+ depends on / blocked
 
Reported: 2014-09-13 15:56 UTC by Richard W.M. Jones
Modified: 2015-04-14 10:34 UTC (History)
5 users (show)

Fixed In Version:
Doc Type: Bug Fix
Doc Text:
Clone Of:
Environment:
Last Closed: 2015-04-14 10:34:52 UTC
Embargoed:

Attachments (Terms of Use)
build.log (2.88 MB, text/plain)
2014-09-13 15:56 UTC, Richard W.M. Jones 		no flags Details
View All

Description Richard W.M. Jones 2014-09-13 15:56:03 UTC 
Created attachment 937243 [details]
build.log

Description of problem:

This is a placeholder for the extremely mysterious heisenbug where
the /dev/sda1 device node does not get created, or magically disappears.

Version-Release number of selected component (if applicable):

It started happening in Fedora 21 and above.

How reproducible:

Rare, but annoying.  Happens a lot more often under Koji than
anywhere else, but can be reproduced in a VM occasionally.

Only appears to happen after we write to the first sector of /dev/sda.

In the attached build log, it happens twice.  Search for the
error message "/dev/sda1: No such file or directory"
Comment 1 Richard W.M. Jones 2014-09-14 11:47:40 UTC 
Ugh.

Turns out to be a stupid udev rule.  Whenever you close a device,
the kernel sends a "change" event to udev which fires off a whole
set of rules.  It runs at least: scsi_id, path_id, blkid.  And these
operations overlap the subsequent mkfs or other operation.

More udev workarounds needed.

Thanks agk for pointing me in the right direction.
Comment 2 Richard W.M. Jones 2014-09-14 13:03:00 UTC 
Added a workaround:

https://github.com/libguestfs/libguestfs/commit/4515413217c7d4a6fe026254bffa6f82746047c4
Note You need to log in before you can comment on or make changes to this bug.