problem installing 6.5.3 in a VCS cluster

[Stefaan_M]

Hi,

current setup:

2-node VCS cluster with Netbackup 6.5(clustered master server)

I want to upgrade to 6.5.3.

Froze NBU group in VCS and installed 6.5.3. on the passive node. -> this works

When I try to install 6.5.3 on the active node it fails.

The CLT binaries are installed but the server part fails with the following error:

_________________________________________________

Running /usr/openv/netbackup/bin/admincmd/bpsyncinfo add_paths command ...
cannot connect on socket

ERROR: Unable to execute bpsyncinfo -add_paths.
------------------------------------------------
Installation of pack NB_6.5.3 FAILED Mon Jan 26 15:29:49 MET 2009 Rev. 1.6.4.14.4.2.
------------------------------------------------
_________________________________________________

Any thoughts?

thanks in advance.

best regards,

Stefaan?.

[Andy_Welburn]

The following is for 6.0 but may be relevant:

GENERAL ERROR: When installing a NetBackup 6.0 Maintenance Pack the error "ERROR: Unable to execute ...

[RAPHAEL_F_CARVA]

You forgot running on active node, after freezing the cluster group:

touch /usr/openv/netbackup/bin/cluster/frozen

You must run this before upgrading.

The whole upgrade procedure is described bellow (page 46/47 of High Availability Guide - http://seer.entsupport.symantec.com/docs/290237.htm :(

To upgrade a NetBackup failover server 

1 Make sure the NetBackup group is online. 

2 Ensure a good backup of your cluster environment exists that includes a 

catalog backup. 

See “Configuring NetBackup catalog backups in a cluster” on page 86. 

3 For each NetBackup server that runs outside of the cluster, ensure that the 

server list is accurate. This list should contain the name of each node on 

which NetBackup can run and the name of the virtual server. 

4 Freeze the NetBackup group using hagrp -freeze nbu_group_name. 

5 Run the following command on the active node to disable the NetBackup 

agent monitor: 

touch /usr/openv/netbackup/bin/cluster/frozen

Note: If the NetBackup resource faults during the upgrade, clear the fault. 

6 On each inactive node to which NetBackup may failover, install the 

NetBackup server software. 

■ Follow the instructions for how to upgrade NetBackup as described in 

the NetBackup Installation Guide. 

■ If required to specify the server name, provide the virtual name of the 

server. 

■ If necessary, refer to the Veritas Infrastructure Core Services 

Installation Guide for details on how to install PBX.

7 On the active node, install the NetBackup server software. 

Follow the instructions for how to upgrade NetBackup as described in the 

NetBackup Installation Guide. 

During an upgrade of a NetBackup in a VCS environment, you see the 

messages that indicate that the database is created twice. These messages 

appear because the database is created in the default location 

(/usr/openv/db/) during install_bp. Later, it is moved or recreated on the 

shared disk. 

8 Unfreeze the NetBackup group using the following command: 

hagrp -unfreeze nbu_group_name

9 Run the following command on the active node to enable the NetBackup 

agent monitor: 

rm -f /usr/openv/netbackup/bin/cluster/frozen

10 Bring the NetBackup group offline and online. 

11 On each node on which NetBackup server is installed, upgrade any 

NetBackup database agents or options. 

Note: You must freeze the NetBackup group before you install any agents or 

options. Refer to “Installing or upgrading NetBackup database agents and 

options” on page 45. 

12 Verify that NetBackup can failover properly in the cluster. Refer to 

“Verifying NetBackup configuration” on page 90. 

[Stefaan_M]

frozen file is there ... I forgot to mention it in my post

resolving looks ok.

(Remark: if I do an nslookup I receive the FQDN instead of just the hostname ... can this be the problem?)

[RAPHAEL_F_CARVA]

Maybe, but before fix that, have a look at bp.conf and ensure that all servers listed there still exist and are available to communicate. 

After this, if everything looks ok on bp.conf, configure /etc/hosts to try fixing nslookup you reported. But I guess your problem is on bp.conf.

[Stefaan_M]

despite my changes to /etc/hosts and checking bp.conf, the problem is still there :(

Any thoughts?

[RAPHAEL_F_CARVA]

try removing from bp.conf any SERVER entrance that refers to a remote console (is not a active media server), like refers to your desktop or operation desktop for example... after this you should bring netbackup cluster group offline then online.

[Stefaan_M]

it's a fresh install so I do not have any remainders of previous consoles...

This is what I found in the postinstall script of the NB_6.5.3 patch:

#  Run bpsyncinfo to create EMM DB Directive for upgrades to 6.0.
#  This directive will update /usr/openv/netbackup/db/config/sync, which is
#  shared in a cluster.  Therefore, this does not need to be executed on
#  the inactive nodes of a cluster.
#
#  In order to run bpsyncinfo command, need to ensure that:
#  --Must be on a master server
#  --Must have dbm running (so don't run this on inactive cluster nodes).

but I do not have a  /usr/openv/netbackup/db/config/sync file

S.

[RAPHAEL_F_CARVA]

Witch procedure have you followed to install this environment? from installation guide or from high availability guide?

To install NBU in a Clustered environment you should use high availability guide (the link is in a previous post).