Hi,

Doug Lytle wrote on 2011-05-21 15:49:57 -0400 [Re: [BackupPC-users] BackupPC on 
XFS getting lots of error -4 when calling...]:
> Holger Parplies wrote:
> > Where do you get the impression there's anything wrong with the file system?
> 
>  >> "2011-05-20 14:39:37 BackupPC_link got error -4 when calling 
> MakeFileLink"
> 
> That's the first thing that came to mind would be to run either an 
> xfs_check on the file system and or xfs_repair.

my point exactly. You shouldn't recommend the first thing that comes to mind
without further consideration whether it actually makes any sense - especially
if it is an operation with the potential to destroy a perfectly functional
file system. I'm not saying xfs_repair will do that, I'm just wary about
repairing file systems. Sometimes you have no other option, sometimes it's
successful. But the very *nature* of repairing file systems is looking for
inconsistencies and *guessing* what the intended state might be. Sometimes
that involves discarding data which might otherwise crash your OS (or the FS
driver), because there is no way to guess the correct meaning of the data.
Other times (like 'reiserfsck --rebuild-tree') you start a process which needs
to complete before you can access *any* data on the FS again. Most of the
time, you are strongly encouraged to make a backup copy of the whole partition
you are about to 'repair', in case the result is not to your satisfaction
(xfs_repair(8), in fact, does not; however, xfs_check(8) describes how to
"save important data before repairing the filesystem with xfs_repair").

So, had you suggested xfs_check, that might have been pointless but harmless,
whereas xfs_repair is pointless and potentially harmful. Well, in my opinion,
at least.


For the archives:
If you suspect your XFS file system is corrupt,

1.) read the man pages xfs_check(8) and xfs_repair(8),
2.) then run xfs_check, and, only if this indicates you should,
3.) correct problems with xfsdump(8) and/or xfs_repair after possibly making
    a backup copy of your file system (i.e. partition/volume).


In the context of BackupPC, you also need to think about

1.) what caused the problem? Can you trust the physical disk for future
    backups?
2.) was the repair action successful? How do you test that for a BackupPC
    pool? Can you be sure no backup is missing files or contains incorrect
    content? Is that important for you? Will you need to know at a later
    point in time, when you access the backups, that they may be inaccurate?

The safe answer to both points is, "start fresh on a new disk", but that is
obviously not always possible.


In the context of this thread, forget about the above. The problem is *not* a
media error, though starting with a fresh pool may be a good idea if you want
to get pooling right. Alternatively, look for Jeffrey's scripts for fixing
pool linking issues.

Regards,
Holger

------------------------------------------------------------------------------
What Every C/C++ and Fortran developer Should Know!
Read this article and learn how Intel has extended the reach of its 
next-generation tools to help Windows* and Linux* C/C++ and Fortran 
developers boost performance applications - including clusters. 
http://p.sf.net/sfu/intel-dev2devmay
_______________________________________________
BackupPC-users mailing list
BackupPC-users AT lists.sourceforge DOT net
List:    https://lists.sourceforge.net/lists/listinfo/backuppc-users
Wiki:    http://backuppc.wiki.sourceforge.net
Project: http://backuppc.sourceforge.net/