Conference decwet::networker

    Hello !
    
    I'm getting the same thing.  I ended up setting the parallelism to 1
    and the sessions per device to 1 as well.  I am running 13 clients with
    the same version.  I am using the default for the savegroups and don't
    understand what you mean about overlaping.  I restarted nsr and it
    appears to be working fine now.  
    
    Steve

re: savegroups overlapping...

  this means that if savegroup #1 starts, and is still running (you see
 saves in the ps output), and savegroup #2 starts up - you now have an
 "overlap". Also, savegroup #2 cannot start until savegroup #1 has finished
 doing the saveindex, *and* media verification, if applicable.

FYI

    Hi,
    
    just to tell you that I am also getting this "device de-activated" with 
    DU V3.2D-1 and NSR V4.2B+P1 with two savegroups ovelapping.
    
    Manu. 

	This is a wild shot but could somebody with the problem set
	the client timeout values to a very high number such as 
	2 hours, this would include the server's own client record.

	Anytime savegroups overlap, each group expects paralellism
	sessions to be available, but since another group is using some
	of the available sessions saves are started but must wait 
	until the other sessions finish.  This could be triping the 
	time out logic a causing less than clean session terminations 
	and getting the session counts out of wack between nsrd and 
	the tape daemons.

    Hi Rogers,
    
    I finally manage o do the test:
    
    NSR V4.2A + P2 & P3 on DU4.0B - 10 Groups all starting at 22:00 -
    
    I change the Timeout to its maximum which is 1000 Minutes, so the
    asavegrp command are displaying on the server as asavegrp -T 60000.
    
    I still have the "tape deactivated during save" message andnsr hangs
    up. I had to unmount and remount the tape to restart the backup.
    
    During my test, I was able to see the message appearing in the message
    window of networker, and surprisingly, the session window cleaned up,
    but the tape was going on and the "throughput/Megabyte saved" were still
    updated in the device window during at least ten minutes before the tape 
    stopped. The asavegrp processes of all the savegroups stayed visible.
    
    Unfortunately, I had to migrate to V4.2B because the version V4.2A
    with patches P1 P2 and P3 was not able to backup "All" the savesets of
    a cluster service I just modified to add a second filesystem. So I took
    the opportunity to change the groups to avoid overlapping of the backup
    time window.
    
    But, in summary two problems are existing:
    
    1) V4.2A and P1 + P2 + P3 + Cluster service + All = only one saveset
       is backup'ed except if you indicate all the saveset in the saveset
       field.
    
    2) V4.2A and overlapping groups and timeout=1000 minutes still give 
       "device deactivated by save"
    
    
    
    Regards, Manu.