offline database backup when package is halted

[TMcB]

Hi everyone

I have configured a new package in our test cluster and it is working fine - it mounts / unmount volumes and starts and stops our application which uses informix database.

On our standalone servers, everynight we shutdown the database for 1 hour to perform our nightly db backups using ontape command.

I'm confused as to how we will be able to do this now in the cluster- if I have a nightly crontab to halt the package - this will unmount the volumes I need to backup using ontape.

Any advice on how to get round this would be greatly appreciated!

[Marcin O.]

Hello

You can always shutdown package, this will unmount volumes, but after that you can run in script mount that volume in exclusive mode
vgchange -a e ..., do backup, deactivate volume, run package.

Marcin

[Steven E. Protter]

Shalom,

Have the backup script issue the same mount command as is in the package control scripts. You should be able to cut and paste, maybe you'll need to set a variable or two.

SEP

[Rita C Workman]

Why umount....can't you just leave them up by using the entries you say is your cntl file that starts/stops the database.

In your nightly cronjobs - couldn't you just have cron run to stop/start the database using that script and leave the mountpoints where they are.

Just a thought,
Rita

[TMcB]

Hi Rita

I'm using the monitor_process function that came with the ECM informix toolkit.
I thought if I stopped informix, it would detect the processes were missing and halt the package.

Thanks

[Rita C Workman · Accepted solution]

Since I don't run Informix, you would be the best to test for that. But is there a way to control via a script so you stop the monitor also as part of the halt database, and then cron that script ?

Just a thought,
Rita

[TMcB]

thanks Rita - that sounds like the answer - stop the monitor for the 1 hour backup period

thanks

[A. Clay Stephenson]

My standard way to do this is to modify the monitor script to detect if a file is present. The backup script creates this file when it starts, the monitor script then detects this file and if found does not do its normal functions but reports "all is well". When the file disappears (because the backup post-exec script removed it), the monitor script resumes its normal operation.
If you do it like this, the backup script doesn't need to halt the package; it just needs to halt the application.

Let me suggest a Plan B which will reduce your 1 hour downtime to minutes.

1) Shutdowm applications/database.
2) Create vxfs snapof mounts or do an lvsplit is using raw data.
3) Restart database/applications.
4) Backup the snapshots.
5) Remove the snapshots.

You have almost all the uptime of a hot backup combined with the warm, fuzzy of a cold backup.

[TMcB]

that is exactly what I'll do - I'll use the temp file to stop the monitoring.

we are using raw volumes for our database - so I'll definatley think a wee bit more into your suggestion for reducing our down time.

You guys have all been a great help!