Advice on Informix Disaster Recovery

[TMcB]

Hello everyone

We have an hpux machine, on which is running Informix Dynamic Server.

We have another hpux box which we would like to use for DR purposes.
Every night after the infomix backup is finished, we automatically copy over the infomix backup files to the spare server.


Could anyone give me some pointers on what would be the best way to be able to copy over each logical log file when they fill up.

Much appreciated!

[Pete Randall · Accepted solution]

You could add something to the logfull.sh script to copy the log to the other machine. Are the machines connected via NFS?


Pete

[TMcB]

Hi
no - we're not using NFS at the minute, but I can set this up. We currently copy over the databse backup files using "rcp"

[Pete Randall]

rcp would do just as well. You just need to add a line to /opt/informix/etc/log_full.sh or whatever you've specified as your ALARMPROGRAM to do the copy. The ALARMPROGRAM will do the backup, then rcp the log to the other system.


Pete

[Rainer von Bongartz]

Another way (that I use) it to restore the database on the second server using the onbar
backup from the production machine.
This way we also verify the daily backup .

Of course this requires an external storage manager for onbar (we use Legato's Networker)

Regards
Rainer

[TMcB]

I've just been having a look at the current setup - the logical logs are backed up once, at the same time as the database backup.
The command used is :
ontape -a.

So to sum up the best way to do this -
change the logical log backups to continuous.
use alarmprogam / log_full.sh
ON the DR machine restore the backup to verify it has worked.

[TMcB]

Just another thought -

would it be better to set up cronjobs eg every 2 hours to run "ontape -a" and copy over the backup file?

Any opinions?

[Pete Randall]

In my opinion, I would prefer the continuous log file backup option. That way Informix tracks what is available and required in the various logs presented to it for recovery. I'm not quite sure how it would handle it otherwise.


Pete

[Steve Lewis]

Here's a different way to do this, using HDR.

Set LBU_PRESERVE=1 in the onconfig file.

Give the backup server a different INFORMIXSERVER name (you can change this easily without restores).
Put the backup INFORMIXSERVER details into the sqlhosts file on the primary.
Put the primary INFORMIXSERVER details into the sqlhosts file on the secondary.
Make sure they can ping each other.

Immediately after level 0, do a physical restore onto the secondary server.
When complete, on the primary server type
onmode -d primary [informixserver of secondary]
then on the backup server immediately type
onmode -d secondary [informixserver of primary].

Monitor the informix logs and you will see it catch up the logical logs over the network automatically and keep you right up-to-date. No dodgy logical log backups to disk (how could you hope to restore them in the right order? +its unsupported)

To invoke DR, it is a simple matter of putting the secondary server into standard mode (onmode -d standard) and then (maybe) bouncing it with the original informixserver name to allow users to connect with no changes. If you allow it to switch automatically it can be a problem.

See the informix DB administrators guide for full details, then test it. Easy.