[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE: [cobalt-users] Did anyone ever successfully restore a Raq4r?
- Subject: RE: [cobalt-users] Did anyone ever successfully restore a Raq4r?
- From: Greg Hewitt-Long <greg@xxxxxxxxxxxxxxxxxxx>
- Date: Wed Apr 11 00:21:08 2001
- List-id: Mailing list for users to share thoughts on Cobalt products. <cobalt-users.list.cobalt.com>
> > Being a bit paranoid I have been attempting to do a disaster recovery of a
>> Raq4 before putting it into production. I have backed up using both the
>> built-in tools both by web and FTP, uploaded the .raq files via FTP to a
>> fresh machine loaded from the recovery CD. The restore only
>> works for maybe
>> half the files and "unknown error" is shown for the rest. Next I took the
>> hack approach and stripped the headers off of the .raq file and
>> restored the
>> tar archive using the same parameters used by the Cobalt CGI,
>> that got back
>> all the missing files but now "Site Management" in the GUI is
>> broken. Seems
>> the site management Perl script is having trouble accessing the Postgres
>> config database (exec failures on sql commands), the database is running
>> though. (Maybe the backup script didn't capture everything needed for
>> recovery???)
>
>well a bit of a confession here the backup and restore process does not work
>in a disaster recovery scenario ( don't flame me i'm only an SE )
>
>> Is a clean disaster recovery possible with the Cobalt
>> backup/restore scheme,
>> is there a trick or something that I am missing?
>
>I think the best scenario I have seen so far is to use the CMU (Cobalt
>Migration Utility) run it as a cron job nightly and then ftp the resulting
>file to somewhere else. If you have to rebuild your RaQ then you can do a
>CMU import to restore the sites,data, users etc.
I've noticed a problem with the CMU - it follows sym-links - so that on restore, what was a sym-link now becomes a copy of the originally linked file. Which is a TOTAL pain in rear for so many sites we have on our servers.
For instance, we have a client site which was developed with 50 user folders, each containing their profile, then it was symbolically linked to their name, their login ID, their first initial and their surname, and a few other configurations of their name.
So, for each user, there are about 6 ways of accessing their profile within the site. On restoring this site from a CMU (I'm moving the box it's on), I'm going to have remove all the duplicate folders, and relink them to the originals. So, for 50 users, there will be 5 rm -r commands, and 5 ln -s commands - ie, 10 commands (yes I can script it).
Whereas I've worked out that removing the h in the -ch of TAR_FLAGS in /usr/lib/cmu/CMU/Archive.pm will probably fix this; can anyone say that they've actually done or tested this already?
thanks
Greg Hewitt-Long
>
>Hope this helps
>
>Gavin
>
>Gavin Nelmes-Crocker
>Systems Engineer, Western EMEA
>SunCobalt
>308 - 314 Kings Road
>Reading
>RG1 4HP
>
>+44 (0)1189 356225 (Tel)
>+44 (0)1189 356210 (Fax)
>+44 (0)7860 181700 (Mob)
>gavin@xxxxxxxxxx
>www.cobalt.com
>
>_______________________________________________
>cobalt-users mailing list
>cobalt-users@xxxxxxxxxxxxxxx
>To Subscribe or Unsubscribe, please go to:
>http://list.cobalt.com/mailman/listinfo/cobalt-users
--
http://www.webyourbusiness.com/
Providers of E-Commerce Software &
Web Design Consultancy and Services.
PH: (970)266-0195 FAX: (970)266-0158