Bug 10899

Summary:	xfs on raid5 hang and serious disk corruption
Product:	File System	Reporter:	dev
Component:	XFS	Assignee:	Christoph Hellwig (hch)
Status:	RESOLVED DOCUMENTED
Severity:	normal	CC:	bunk, hch
Priority:	P1
Hardware:	All
OS:	Linux
Kernel Version:	2.6.25	Subsystem:
Regression:	Yes	Bisected commit-id:
Attachments:	Crash log and following boot

Description dev 2008-06-12 08:19:37 UTC

Latest working kernel version: 2.6.24
Hardware Environment: x86_64 AMD Athlon(tm) 64 Processor 3500+ AuthenticAMD 
Software Environment: Gentoo GNU/Linux

Problem Description: Sudden system hang (hard lock, magic sysreq not working), after reboot out of sync raid5, after resync xfs_check failed because of corrupt superblock, xfs_repair restored superblock with secondary copy, but detected open transfer in metalog, suggested mount failed because of missing root node. repair succeeded with xfs_repair -L
End of day: Nothing left on disk ca. 360Gb out of 600Gb in lost+found most of it usable. System log before hang and initial boot afterwards attached. If anyone needs more information, feel free to ask( sry faq is down at the moment)

Steps to reproduce: Thanks, do not even want to try!

Comment 1 dev 2008-06-12 08:28:08 UTC

Created attachment 16462 [details]
Crash log and following boot

Comment 2 Christoph Hellwig 2010-02-09 22:26:35 UTC

This was with disk write caches enabled, and raid code that didn't support barriers.  I suspect you simply lost the content of all disk write caches.

Comment 3 dev 2010-02-10 10:32:04 UTC

As never happened again since, I'll just leave it closed but i still think that it is kind of funny for it to loose content that had not been touched for month.