XFS Repair hangs at inode repair in phase3

[Dave Chinner]

[ A couple of email-to-lists tips: Please:

	- don't post in html format, use plain text
	- please don't top post, commentin line
	- fix the in-reply-to fields on your email replies
	  so that mail programs can thread the conversion properly
]

On Thu, Jun 26, 2014 at 09:17:27PM +0200, Dragon wrote:
> i upgraded to debian jessie with xfsprogs 3.2. if i run xfs_repair
> /dev/md2 the system hangs too. If i use -P it ends up with:
> corrupt block 0 in directory inode 2147702899: junking block
> Segmentation fault

Can you run this under a gdb and get a stack trace from where it
crashed. You might need to grab the source and build that to get a
meaningful stack trace....

> No modify flag set, skipping phase 5
> Inode allocation btrees are too corrupted, skipping phases 6 and 7
> No modify flag set, skipping filesystem flush and exiting.

What actually went wrong with your storage? The only time I've seen
that warning is when a raid array had been reconstructed incorrectly
after a series of disk failures. Did your RAID have failures or
reconstruction problems before XFS started reporting errors?

Cheers,

Dave.
-- 
Dave Chinner
david at fromorbit.com