Re: Recovering from disk errors

On Friday 02 July 2004 21:43, Ed Leafe wrote:
> 	I'm hosed. I woke up this morning to a series of I/O error messages on
> my server. They were so bad that I couldn't even log in at the console
> to try to correct the problem.
>
> 	I rebooted, and ran the drives through the SCSI utilities to check for
> problems, and then fsck'ed like crazy, due to all the problems
> reported. Supposedly, though, the disk is fixed, at least to fsck's
> liking.
>
> 	I can't start Zope, though. Since I store my data in /usr/zopestorage,
> and the highest number of errors was in the /usr partition, I'm not too
> surprised. I looked at the docs on the website, and found "checkds.py -
> DirectoryStorage checking tool". Sounds like a good start, so I ran it:
>
> [root@leafe DirectoryStorage]python snapshot.py /usr/zopestorage -
> python checkds.py
> failed to open /usr/zopestorage/config/snapshot.conf
> SnapshotError: directory is not a snapshot
>
> 	OK, not too surprising. So I renamed this to zopestorage.BAD, and then
> restored from my backup. But when I ran it again, I got the exact same
> result.
>
> 	So what do you suggest I do next?

You still get the error message 'failed to 
open /usr/zopestorage/config/snapshot.conf'? Does that file exist in your 
restored backup?

Hmmmm.... Ive just noticed a gap in the documentation for backup.py which 
explains this, assuming you used backup.py to take your backup. That tool 
only takes backup copied of the storage *data*, the files in the 
directory /usr/zopestorage/A. It doesnt backup /usr/zopestorage/config, or 
any of the other directories in /usr/zopestorage/

No problem - all those files are easy to recreate. I suggest:
1. move your restored /usr/zopestorage/A into a safe place. This contains
   all your data.
2. run mkds.py to create a new storage.
3. remove the newly created /usr/zopestorage/A, and replace it with
   your restored one from step 1
4. recreate your config.snapshot.conf. (ok, I lied. all but one of these
   files are easy to recreate ;-)

mkds.py wont be doing anything complicated for you here. If you prefer a more 
hands-on approach:
1. cd /usr/zopestorage
2. mkdir journal
3. mkdir B
4. put the default settings file in config/settings
5. put a random number in config/identity. On linux,
   /proc/sys/kernel/random/uuid is a good source. The only purpose of
   this identity is replication.



-------------------------------------------------------
This SF.Net email sponsored by Black Hat Briefings & Training.
Attend Black Hat Briefings & Training, Las Vegas July 24-29 - 
digital self defense, top technical experts, no vendor pitches, 
unmatched networking opportunities. Visit www.blackhat.com