Preferred method for backup and disaster recovery
What methods do you have in place for a complete backup and disaster recovery of an OTRS system? What is the preferred method to backing up the system both in circumstances with and without a tape drive in the machine itself? If the user who ultimately might have to restore your system is not a native Linux user, but does use Unix for other systems, would you recommend anything different? It would appear the obvious solution would be a mysql dump and grabbing the Config.pm file, but there must be more to it than that, including a more transparent solution from disaster to recovery. Thank you!
On 5-04-2005 at 20:14, Cainkar, Paul wrote:
What methods do you have in place for a complete backup and disaster recovery of an OTRS system? What is the preferred method to backing up the system both in circumstances with and without a tape drive in the machine itself?
We do an incremental rsync of /opt/otrs and Apache directories on a off-site box, so that we can also keep old copies. Disaster recovery includes not only backing up data but also being able to quickly reinstall applications without having to work on their compilation, configuration, permission settings etc. Furthermore, the meaning of disaster recovery in your specific environment depends on hardware configuration (RAID?), needed timing (30 min? 60 min? 2 hours?). For example in a mission-critical setup you would have a read-only disk image of the OS + Apache + /opt/otrs and then a disk for storage: this way you don't have to worry about your startup disk and only get images of the storage disk. In a cluster setup you may have a NFS export with application and data or even PXE boot. As always, It Depends(tm) ;) - Alessandro Ranellucci
Hello! In my case I don't need external backup (yet), so I use two servers (one P4 1.5 Mhz and one P3 866Mhz, both with 320Mb, very basic hardware). Both are configured in High Availability with heartbeat + DRBD. This servers don't have any other service yet. Very basically, they share a partition called /HA, and I have symlinked /usr/share/otrs to /HA/otrs and there's also a symlink in /var/lib/mysql to /HA/mysql so that mysql DB's are replicated. The also share a floating IP used to access to OTRS. Apache virtual host is also shared, otrs crontabs too, etc... Both servers are updated at the same time (Debian apt-get ;) to make them virtually identical, at least in software. If my main server (P4) goes down, heartbeat takes care of it, and starts the services (drbd, mysql, apache2) in my backup server (P3). It takes a bit more than 3 minutes to have all the services completely restored (I'm using a custom apache2-perl- startup.pl to "compile" OTRS at startup time, the app is a LOT faster this way but it takes about 2:40 minutes to startup!!). On the other hand, I have a sligthly modified version of the backup script included with OTRS, adding more files (Crontabs, apache configs, article directory, the whole Kernel dir...). Those backups are copied via scp once a week to an external server, so I'll have something if I lose completely my servers (stolen, burned, fried or crushed by Godzilla XD). It's a flexible setup, that allows me to run the application on one server while putting the other in maintenance, with very low impact in the service, even in case of hardware failure. Something like "application and data RAID 1 over the network" ;) Regards, On 5 Apr 2005 at 12:14, Cainkar, Paul wrote:
What methods do you have in place for a complete backup and disaster recovery of an OTRS system? What is the preferred method to backing up the system both in circumstances with and without a tape drive in the machine itself? If the user who ultimately might have to restore your system is not a native Linux user, but does use Unix for other systems, would you recommend anything different?
It would appear the obvious solution would be a mysql dump and grabbing the Config.pm file, but there must be more to it than that, including a more transparent solution from disaster to recovery.
Thank you!
--- Victor R. Rodriguez Departamento de Sistemas Valoraciones del Mediterraneo, S.A. ---
On 7-04-2005 at 12:24, Victor Rodriguez Cortes wrote:
(I'm using a custom apache2-perl- startup.pl to "compile" OTRS at startup time, the app is a LOT faster this way but it takes about 2:40 minutes to startup!!)
Oh, would you mind to share it, after having appetized us so much? ;-))) - Alessandro Ranellucci
Of course! But it's nothing new, it's a standard feature of mod_perl (that's what I think, maybe I'm wrong!). Take a look at $OTRSHOME/doc/"README.webserver" and to $OTRSHOME/scripts/apache*.pl. If you're running apache2+mod_perl2, please read this post to avoid some weird problems: http://lists.otrs.org/pipermail/otrs/2005-February/007379.html What you're doing with this is to make a static connection to MySQL that will be used instead of connect and dissconect each time you make a query in OTRS and compiling at startup all the modules you specify in your startup.pl file. The difference is awesome: with my little server a queue view with 12 tickets took about 3 seconds to render before, and now i think it's less than 0'3 seconds :) Try it, it's really worth the time! Regards, On 7 Apr 2005 at 12:58, Alessandro Ranellucci wrote:
On 7-04-2005 at 12:24, Victor Rodriguez Cortes wrote:
(I'm using a custom apache2-perl- startup.pl to "compile" OTRS at startup time, the app is a LOT faster this way but it takes about 2:40 minutes to startup!!)
Oh, would you mind to share it, after having appetized us so much? ;-)))
- Alessandro Ranellucci
--- Victor R. Rodriguez Departamento de Sistemas Valoraciones del Mediterraneo, S.A. ---
participants (3)
-
Alessandro Ranellucci -
Cainkar, Paul -
Victor Rodriguez Cortes