xfs_repair

[Eric Sandeen]

On 11/14/13, 11:51 AM, Roman Hlynovskiy wrote:
> hello, after a server crush we are trying to get back access to data on another server.
> 
> out setup is the following:
> 4 disks in md-array, lvm over it and xfs
> 
> xfs_check just stops with 'out of memory error'

xfs_check is deprecated & doesn't scale; don't bother.
Use xfs_repair -n if you want a readonly check.

> xfs_repair after a long list of complains says:
> 
> corrupt dinode 2473604576, extent total = 1, nblocks = 0.  This is a bug.
> Please capture the filesystem metadata with xfs_metadump and
> report it to xfs at oss.sgi.com <mailto:xfs at oss.sgi.com>.
> cache_node_purge: refcount was 1, not zero (node=0xb43ab688)
> 
> fatal error -- 117 - couldn't iget disconnected inode

Are you using latest xfsprogs?

> I captured this metadata file, but it's 2GB size.
> is there a chance to fix the fs?

how big is it if you compress it?

-Eric