Legato Networker with pre commands

[Wayne Willess]

“First the problem.”
We are in the middle of implementing Legato Networker to backup our HP-UX systems. I am trying to run a pre command script that will stop an application that runs on the server before we back the server up. The problem is that when the script fails Networker will execute the script a many times as the number of filesystems the client is scheduled to backup times the number of retries configured for the client. “i.e. 25 filesystems + (1 retry * 25 filesystems) = 50” I have been in contact with Legato support and they are saying it “is behaving as designed”. I need the script to run once and if it returns an error (my scripts handle errors) fail the client then, not after 50 failures of the pre command script.

“Finally the question.”
Is anyone using Legato Networker with pre and post commands? If so are you aware of this “feature”? What if anything are you doing to correct it? Does this design make sense to anyone? Why would you want this to happen if upon the first successful completion of the script it is not ran again?

I will assign points for responses.

[Pete Randall]

Wayne,

This answer is going to be of any help to you, but . . . . .

This is exactly why I converted from Networker to Omniback a few years ago. I simply got tired of talking to support folks I either couldn't understand or disagreed with. Despite the fact that I actually liked Networker better, life has been much better since.


Pete

[Leif Halvarsson_2]

Hi,

If you can't change the behaviour of Networker you perhaps can change rhe behaviour of the script. If the script fails it creates a flag in a tmp directory. Next time the script is started it checks for the flag and exits if the flag exists. After the backup has run, the flag should be removed, perhaps with a cron job.

[Wayne Willess]

Thanks Leif, I have already writen the script to check for a lock file. If the file does not exist create it and continue. If it does exist exit the script. The lock file is only removed if the script completes successfully or by some one removes it after they correct the origal problem. I really dont like this solution so I was looking or more options and was wanting to know if anyone has gotten a different response from legato on this issue.

[Jordan Bean]

About a year ago I wrote an pre/post-processing script with exhaustive error handling to perform EMC TimeFinder backups of live Oracle databases through NetWorker for my previous employer. If you'd like the see the script and configurations, I can put you in touch with their Operations Manager. Contact me at jbean(at)symark(dot)com.

The "feature" was avoided by archiving the parent directory in which all the file systems are mounted. If the parent is /, then you've got a real problem, unless you want the whole system. The chosen mount points look like this: /bcv//. For example: /bcv/dgWPR1/ora01. The backup group lists only /bcv/dgWPR1 and the backup command includes the switch necessary to traverse mount points (thus eliminating the need to specify each file system separately in the backup group).

[David Nardi]

I sympathize with your suffering of Legato support. I have been fighting with it for years, but have not jumped ship as Networker is performing as needed. But, 2005 may be the year...

What version are you using? I've been using pre & post commands successfully for years (primarily to close databases and backup their file systems cold, or to move directives into place).

If not setup this way already, try changing the saveset to "ALL" in the client setup. Then, for specific file systems to skip, use a .nsr (directive) file in the root directory (or each directory individually).

Good luck~

[Wayne Willess]

David,

We are using Networker 7.1.2. My pre & post commands will do one of two things.

1) Stop an SAP system on a remote server, then split an EMC TimeFinder device group and mount the devices to our backup server to be backed up.

2) Put the Oracle database tables in backup mode on a remote server, then split an EMC TimeFinder device group, return the remote database tables to online mode and mount the devices to our backup server to be backed up.

You can see why I do not want to run this pre command more than once. Have you every tested what Network will do if your pre scripts returned an error?

Thanks

[Alzhy]

Wayne,
We are also using Networker albeit on our Solaris Systems. We've NEVER used the pre/post features of Networker. Instead, we have separate scripts to stop/suspend and application/DB instance and resume them after our split mirrors our done. The NETWORKER job, also scripted via "savegroup" command and not run via Leagto's own scheduler, checks for success/failure status on the preceding scripts.

I think this will be your best avenue. If you need to strictly use Networker's scheduler and use the pre/post scripts.. then you need to be absolutely sure your scripts are well written to make sure Networker is informed of its success or failure -- which will dictate whether to continue with whatever it needs to do or not.

[Jordan Bean]

I've been given permission to reproduce my work here. The attached zip file has my two scripts and simple setup paper.

[Wayne Willess]

Thanks to all who responded. We have completed most of the migration to Networker. I have abandoned the prescript feature and gone with running my scripts from cron. All in all cron is probably the best way to go as it eliminates the change of an operator starting the wrong backup and shutting down a system.

Thanks Again,

Wayne

[Rathin]

Hi,
I am facing the same problem during failure of script. Recently we implemented Networker 7.1.1 on HP-UX 11.11. We do server-free backup (backup from SAN BCV mirror). Some scripts were written database hot backup /cold backup as well as to split the BCV. Problems happens while BCV splitting fails. After failure the same script retries many more times. We are continuously communicating with Legto-Support. We didn't have a satisfactory reply yet. Looking for help