02-08-2008 03:28 AM
I am having trouble with a fail over process to a replacement DR master server.
I have a ‘poor man’s cluster’ set up, with 2 master servers, Live and DR. /opt/openv and /indexes are on mounted file systems.
The fail over process is to dismount these file systems from the Live master server, unplumb the VIP and virtual interface (nbkpmstr) , mount the file systems on the other server and plumb the VIP and virtual interface.
PBX has been installed on the DR server prior to this operation.
This operation has worked fine several times in the past, however I have had to upgrade the box that the DR master server is running on.
My UNIX SA’s say that the new box is exactly the same as the old one and the /etc/hosts file and routing table is correct.
Using the bpclntcmd command it seems everything is connecting OK.
Once the fail over operation is complete, I can fire up the Java GUI, can see all my media servers, can see all my images, however when I submit a restore if fails with an ‘error 25 cannot connect on socket’.
When I submit a manual backup it does not start.
If I tail the /var/adm/messages log, it appears the nbjm dies as the restore starts.
I have tried restarting PBX, and it all seems to be running OK when I do a bpps –x
I have called out Symantec about this, however the last engineer I spoke to told me this fail over method was not supported, but I have been assured it is by other people, and I was running fine before.
I am running NBU 6MP5 on the master server and all media servers. The OS is Solaris 8.
I am reaching the end of my knowledge, so any help would be approached.
Many thanks John
02-08-2008 07:00 AM