Data recovery

[jbk at mail2.gis.net (jbk)]

James R. Van Zandt wrote:
> James Kramer wrote:
> 
>>And don't use fsck. Use e2fsck for an ext3 system. Fsck is what comes
>>up first when your system crashes, but say no, escape, reboot with
>>rescue disk and use the afor mentioned e2fsck.
> 
> 
> I see this in the fsck man page:
> 
>     In actuality, fsck is simply a front-end for the various file
>     system checkers (fsck.fstype) available under Linux.  The file
>     system-specific checker is searched for in /sbin first, then in
>     /etc/fs and /etc, and finally in the directories listed in the
>     PATH environment variable.  Please see the file system-specific
>     checker manual pages for further details.
> 
> also I note       
> 
>   vanzandt:/var/mail $ ls -il /sbin/*fsck*|sort
>   57966599 -rwxr-xr-x  3 root root 130152 Aug 22 00:55 /sbin/e2fsck
>   57966599 -rwxr-xr-x  3 root root 130152 Aug 22 00:55 /sbin/fsck.ext2
>   57966599 -rwxr-xr-x  3 root root 130152 Aug 22 00:55 /sbin/fsck.ext3
>   57966606 -rwxr-xr-x  1 root root  18208 Aug 22 00:55 /sbin/fsck
>   ...
> 
> so e2fsck is actually the same file as fsck.ext3, which will be called
> by fsck for an ext3 filesystem.  Are you saying that ignoring the
> journal improves the filesystem check?  I would expect that to
> increase the number of inconsistencies in the metadata.
> 
>                   - Jim Van Zandt
> _______________________________________________
> Discuss mailing list
> Discuss at blu.org
> http://olduvai.blu.org/mailman/listinfo/discuss
> 

I have had two types of experiences with system craches. The 
first type the automatic journal recovery succeeds and the 
boot continues as normal. In the second, the journal 
recovery fails and you are dumped to a prompt where you have 
to decide what tool to use. From this point I have had 
disastrous results using fsck without rebooting with a 
rescue disk.

What is the procedure you use? I, fortunately have never had 
a crash on my server that wasn't recovered by the journal, 
only on my laptop, which is the test bed for new 
configurations.

till next
	Jim Kelly-Rand