Disaster Recovery

Hi Gary,

It depends on how you do your recovery. The simplest way is to build a box, install Cloverleaf, recreate your sites using siteinit and then restore the site directories from backup. I tend to backup just my sites for this purpose.

I also do full backups of the box but I find this method easier. You just have to ensure you copy a site after any changes.

Regards

Garry

Gary.

If we had to use our dev server if we lost the prod (AIX HACMP) servers, we would execute the same thing

1. create the site

2. install our backup, which is taken from prod on a site basis every week.

This is the main way we migrate as well.

Gary,

One thing to remember is that you mustn’t backup the exec directory as this will be recreated for you by siteinit and at runtime.

Regards

Garry

Restoring a back up at another site is a decent plan, but there is one thing you may want to consider.

If there were any messages stuck in the engine during the disaster, which could be a handful if all systems were go, or a bunch if a receiving system was down, you need to make sure you recover those messages prior to turning on the normal feeds, especially on those threads where the order of messages received is critical.

In the case of recovering a system from a backup, these messages are not present in the recovery database.  This recovery by hand is not impossible, but can be time consuming.  It involves comparing what was last received by each recipient system vs. what was sent from the senders.

In the case of a clustered system with shared disks (DASD or SAN), the recovery database comes over to the fail over server and no messages are lost.  No manual recovery is needed.

Hope this helps,

Great topic – quick question:

What specific files/directories should be backed up?

If you do come up with a list of directories and/or files to backup I’ve been finding the rdist command under AIX very useful for keeping a copy of directories and/or files in sync across platforms.