Monthly Backup

[podsds]

I was asked today if Netbackup has the feature to store the last valid backup per worstation for more than one month.

Unix 3.4 master & media
MT & W2K workstations Client.

I know I can duplicate my policies, change the schedules and the retention period for a month so that the last tape stays around longer. This may work for successful month end backups, but I also need to accommodate failed month end backups and hold the last good backup that occurred.

Anybody ever do this?

 

[comtec17]

There is no built in method for doing this. You would have to write a script for that. Just as an FYI...NetBackup will not save a backup that has failed unless you are using Multiplexing/Multistreaming.

 

[podsds]

I thought this would be a script, okay nothing new there.

I did learn something ... can you tell me more about Multiplexing/Multistreaming and failed backups. At my previous gig we used Multiplexing/Multistreaming together for every backup, but at my new gig they only use Multistreaming ... so I guess the question I am asking now is what does netbackup do with a failed Multistreaming backup.

Thanks for info.

 

[comtec17]

Ok, here is goes...If you multistream, your files piece of the class should look like something like this:

NEW_STREAM
C:NEW_STREAM
D:\users
NEW_STREAM
D:\files

and so on. Each stream is considered a backup job.

So now lets say the C:\ stream completes and the D:\users stream completes but the D:\ files stream fails. The C:\ and D:\users streams are completed and succesfull so these backups will stay on the media for the retention period specified. Now, lets say the D:\files stream failed 50% of the way through and wasnt restarted for some reason. The D:\files stream is failed and the database will not retain the information backed up.

Hope this helps.

 

[podsds]

I can understand that, but isn't there a mechanism that creates streams for you by doing some analysis.
Its probably the same as doing it manually. In my opinion Netbackup need to mature a bit. I have often told Veriats to implement a backup header and footer process to unify all the streams, but they either don't get it, don't care or have other priorities.

 

[comtec17]

This function is very usefull. I used it on all my large servers to split data backups. I had 68 server doing multistreaming due to them have 200GB + disk space used. I created 10 streams of around 20 GB a piece for my data drive. This helped me out because if the backup failed after 180GB backed up, it would only fail on 2 streams and I can restart the backup and it would only backup the streams that failed.

Very usefull in my mind.

 

[podsds]

So you would restart the whole backup and it, Netbackup, would only backup the previous two streams that failed. This happens automagically or do you start a backup only for the two streams that failed. Is this a hot or cold backup?

I am not arguing that the feature is not useful, my gripe is that Veriatas needs to unify all streams, but this is another issue.

 

[comtec17]

You would restart the entire job and it will no that the others have backed up.