Unsolved
This post is more than 5 years old
198 Posts
0
1272
NetWorker 7.6 with Veritas Cluster Server on Solaris 10
Have issue with fresh install of NetWorker 7.6 on Solaris 10 with VCS.
The cluster presents the IP and the shared storage to the cluster nodes, but currently not managing the application. The NetWorker.server command has been used to set up the cluster environment. “/usr/sbin/networker.cluster” to update the configuration –
Prompt |
Response |
Directory where local NetWorker database is installed |
/nsr |
Configure for both NetWorker server and client |
Yes |
Add the site specific values for NSR_SHARED_DISK and NSR_SERVICE_ID in /usr/sbin/nw_vcs |
Yes |
In what path will the shared nsr directory be created/located |
/nsr_shared |
Enter the Logical Hostname to be used by NetWorker |
cmsbk01-vv02 |
However, when the application is started from the command line "nw_vcs start" the following error message is reported -
can’t start nsrd because %s (/nsr) is local, and networker is configured as a cluster server. Use cluster manager to check NetWorker service status. 1 23 8 /nsr/res
A quick "nudge" as to why this is happening would be appreciated.
Thanks in advance
IKP
Have now resolved what is causing the issue - not resolved it! The VCS resource group for the NetWorker server requires two virtula network interfaces - one for the application and the second to allow routing between the cluster a number of clients. However, when there are two IP addresses in the group NetWorker fails with the above error. If the "routing" ip address is removed, then all is OK and the cluster works - except a number of clients are unable to access the virtual resource.
Message was edited by: IKP
R_Friberg
34 Posts
1
June 12th, 2010 09:00
Hi,
We also had this problem although we're at 7.5 sp2. EMC probably wants you to set up the NWClient resource in the cluster group if you have several NICs/IP addresses in the cluster. We have our backup server connected to four different subnets, so it didn't work out of the box for us. It doesn't make sense to use NWClient resource on a backup server if you just want all your NICs to be used for different subnets. The NWClient resource is used to map an IP address to what disks are owned by it.
We ended up with modifying the lcmap script which makes the check that there is only one IP address for the service group where NetWorker is installed. Now we don't have it query the cluster for the nr of IP addresses anymore and that seems to do the trick for us.
Thanks
Rickard
IKP1
198 Posts
0
June 15th, 2010 09:00
Hi,
Have finally managed to get the cluster operational, the issue appears to be in the published monitor process "/usr/sbin/nw_vcs monotor". Have removed this from the cluster resource specification and now the cluser is operational.
regards
IKP