Video Screencast Help

DAG preferred node not working as it used to...

Created: 14 Oct 2013 | 4 comments

DAG has 2 passive nodes for backup. node1 is a media server, node 2 isnt.

Hence preferred list is node1,node2.

Historically if the backup had issues with a db on node1 it would revert to node 2. When node 1 was healthy, one had to remove node2 from the preferred servers list to force all backups back to node1. It worked.

Now Ive got a new scenario: node2 was being permanently used for whatever reason, so I removed it from preferred servers and the overnight backup worked.

I now add back in node 2 and the backup has now reverted to node 2, not what I want. 

Any ideas? Maybe it distinguishes between full and incrmental : i am fairly sur eboth he full and the incr ran on node 2.

The BUG I have with 'preferred servers' is it isnt honoured consistently, it gets overridden by "what worked last time" according to my observations.

Im on 7504 solaris master.


Operating Systems:

Comments 4 CommentsJump to latest comment

RamNagalla's picture

HI Jim,

its working as expected... 

it looks for the last successfull server and check if that is listed in prefer server list ( does not consider the possition on the prefer server list) if it is , it always pick the same...  as you have specified that as one of the prefer servers.

see the page number 112 in Netbackup exchage admin guide.

Backup statusforExchangeDatabaseAvailabilityGroups
(DAGs) andthepreferredserverlist
In anExchange2010DatabaseAvailabilityGroups(DAG)environment,NetBackup
tracks thesuccessorfailureofabackupattemptforeachpassivecopydatabase
backup. ThisinformationisstoredintheBackupStatusdatabaseontheNetBackup
master server.Thisstatusisusedinsubsequentbackupattemptsforeachdatabase
in aDAG.IthelpsdeterminefromwhichExchangenodetoperformapassivecopy
database backup.
NetBackup choosesaserverfromthepreferredserverlistforasubsequentpassive
copy databasebackupattemptasfollows:
 If the last backup was successful and the last backup server exists in the preferred server list...
NetBackup uses that sameserver.
jim dalton's picture

So as I say its not working as expected on two counts:

- the prefererred servers list has an implied order since you can move a server up or down the list.

-as I said I removed node 2 from the list  and it used node1 successfully for the whole DAG: node 1 is thus the last and successful and existing. But I added node2 back in the list and it used node2.Node 1 was still running and was the last successful.

So unless Ive misread your input Nagalla, the reality and your statements dont add up, sorry!


RamNagalla's picture

it gets overridden by "what worked last time" according to my observations.

by this statement i was in a impression that that the node 2 was the last successfull host...

did you checking the status of the backup in Database...? does it also shoiwng node 1 ?

 command :- bpclient -All-EXDB

you can also try resetting the backup status in database and see how it works

bpclient -client host_name -update
-exdb <db_name:server_name[:timestamp:count:status]>
jim dalton's picture

Last good was on node1, I added node2 back in to the list, off it goes to node2, contrary to what used to happen and contrary to my understanding.

I'll explore the bpclient option you mention Nagalla, its not one Ive investigated before.