I want to schedule a GROUP and configure it with a post-exec script, this script will be executed at the end of the backup starting the next group and so on for the following groups.
What you wish is sort of cascaded backups. Out of the box, you can't do it, but since you can virtually configure anything in NW there is a way. One way of doing it is to have master group with some dummy saveset for client. This client would run savepnpc which would execute groups one by one as soon as previous one completes.
You could also write a script on your server that would check for running savegrp processes and run the next savegrp command from the list once the process is finished. I haven't tested it but I don't see why it wouldn't work. That way, you could probably build in extra functionality to deal with a group that is stopped (waiting for a tape for example) so that it wouldn't stop other groups from running.
Assume you have client with 4 savesets: C: D: E: F: defined. You can set the parralellism (client properties - Globals (1 of 2)) to 1, in the client, or in the group, or depending on your environment - number of drives - you can set the max sessions (or target sessions).
Then only one of these will run simultaneuosly. It may not be in the order you wish, however, I still fail to see the reasoning behind why/what you want to achieve. If you can give reasons why you want your exact method, then we can discuss/ help.
I have many Exchange clients with 4 savesets. I have set the client parrallellism to 3, so all SGs do not run simultaneously. Once the first one finishes, the last one kicks off. Would possibly make more sense for me to set to 2, so that the first 2 would run, both would finish similar times, then the last 2 would run???
Phil
when I first started NetWorker, I used to think a "group" as a like minded set of computers, but it is not. A group is a collection of client that start at the same time - much to my dissapointment I used to think this should be called schedule. Also a group helps direct which tape pool to use. "Schedule" is defined more as what level to do, (as opposed to when to start), so I can understand that the terms are not intuitive, even after nearly 10 years of NetWorker.
I'm in the process of implementing cascaded backup for a number of hosts (one IBM HTTP server and a couple of WebSphere process servers) . Here is what I have already done. I deployed two sets of probes. The 1st set of probes shuts applications down prior to backup and the second set of probes kicks applications off after backup. For each host I want to backup I have created 3 NetWorker clients: 1st client have have "shutdown" probe associated with it and backs up some folder. The second client have real saveset I want to backup associated with it and, finally 3rd client invokes "kick-off" probe and back up some folder. Next I created 3 Networker groups. The first group named something like foo_stop incorporates 3 clients with "shutdown" probes. The second group, foo_backup, has 3 clients with actual savesets bount to it. And the 3rd group have 3 clients with "kick off" probes associated with it.
After that, I have tinkered a script that runs these 3 groups consequently. It looks similar as follows:
@echo off
set LogFile=nsrprobe_peprod.log
echo "Script started..." >> %LogFile%
rem Initially run group to stop applications on all clients
rem First Stop IBM HTTP Server
savegrp -c httpserver -G foo_Stop
if %ERRORLEVEL% NEQ 0 goto :FAIL
echo "httpserver stopped..." >> %LogFile%
< >
rem Now invoke foo_backup group to backup all clients
rem First backup IBM HTTP Server
rem savegrp -c httpserver -G foo_Backup
rem if %ERRORLEVEL% NEQ 0 goto :FAIL
< >
rem Finally bring all applications online by using
rem foo_Start group
rem start all WebSphere Process Servers
savegrp -c websphere1 -G foo_Start
if %ERRORLEVEL% NEQ 0 goto :FAIL
echo "websphere1 started..." >> %LogFile%
<< Next I start HTTP server >>
rem If the program flow managed to get here then our mission is accomplished
exit 0
:FAIL
exit 1
I put that script on my Windows-based NetWorker server. Next, I have invoked that script a few times to make sure that it works fine. As I launch that script I, see how groups and probes are invoked in a correct order.
Finally, to put this rather complicated construction in action I have created yet another group named foo. For this group I configured savepnpc backup program and created foo.res file that has the bat file listed above as precmd.
The problem is once I kick that foo group off, the precmd script, in turn, kicks foo_stop group off and then the whole backup process gets standstill. The foo group turns to be in the running state whereas foo_stop group is in probing state.
guenterH1
117 Posts
1368
0
Posted June 22nd, 2010 23:00
Hello,
I'm not sure whether I understood what you mean with 'concatenate backups'. In NetWorker this term is not known.
Probably an example from other backup software would help to clarify what steps you mean.
However NetWorker knows about a backup level consolidate.
This backup method uses a previous level full backup and a new level 1 (differential) backup which will be merged together to a new full backup.
This method allows you just to make an initial full backup of your client and then subsequently only differential backups.
For the last differential is merged to a new full the usual effect differential growing as time goes on is avoided.
However this backup level consolidate has some limitations, please read the documentation about this carefully.
Probably this is what you're looking for.