Upgrade from 6.5.1 to 6.5.3

[subhojit_pal]

Hi Folks,

Netbackup Platform: 6.5.1
OS: HP-UX 11.11

We have one master server which is 6.5.1 and runs around 6000 - 6500 jobs daily and backup approx 40 TB daily. Before upgrading I had checked all the lernel parameters as per installation guidelines and found all of them to be more than what has been recommended.
After upgrading from 5.1 MP5 to 6.5.1 we found three major issues.
    1) the nbpem either dies or core dumps every couple days.
    2) the NB_dbsrv process takes a lot of CPU.(Almost always 100%)
    3) it is taking a long time to dismount/ mount tapes during peak backup window.

Symantec suggested to upgrade Netbackup to 6.5.3. Have any one has faced similar issues? If yes, has upgrading to 6.5.3 resolved the issue.

I also heard from one of my ex-colleague that in his setup(Solaris 10), the server crashes every couple of days after upgrading to 6.5.3. Has anyone else had seen this issue.

Thanks,

Subhojit.

[CY]

I don't have NetBackup Master/Media server on HP-UX now a days.

There are some bugs fixed related with HP-UX platform from 6.5.1 to 6.5.3, but you will still need a new bpbkar binary if your HP-UX server has certain OS patch - see this bug report:

http://seer.entsupport.symantec.com/docs/313359.htm

We have 6.5.3 on Solaris 10, and we do have some problem with it but I don't know if it's the same problem your ex-colleague have.  Per the Symantec support engineer, our problem is due to Sybase DB engine which Symantec is working with Sybase to solve it.

Ask the support engineer if 6.5.3 would address the issues you have.  If they can't answer, escalate the case and get a back-line engineer to engage.

[Anonymous]

My master is a HP-UX 11.23 (IA64) server.
Not directly experienced your issues but heres my background

6.5.1 updated to 6.5.2
introduced a NOM server for the first time
- now Master Server cores repeatedly, when NOM server poles the master for large data requests it apparently cores
- Support Case told me to turn off the NOM server for good until new update release
- also to look through policies for any files lying in the policy folder that should not be there

6.5.2 > 6.5.3
within 30 minutes of starting up - core file produced, problem initiated around the vmdareq command or use of the Device Monitor. I have several media servers in a SSO configuration, and there netbackup processes all froze and couldnt communicate with the master.

6.5.3 > 6.5.3.1
core again. Analysing the core it is coming from nbemm

But have never had a core for over 3 weeks now.

No fix forthwith from Symantec. Although there is this recent technote
http://seer.entsupport.symantec.com/docs/313880.htm
But infers PA-RISC

Anyway in 1 months time I'm jumping off the HP-UX gravy train to a new SLES 10 Master server x86 architecture.

My comments would be to go to the latest updates, they have fixed issues in the past, and its not like you can ideally go back.

The problem with nbpem has been well documented and have been fixes ongoing for this through applying 6.5.2 and 6.5.3

Anyway there looks like the release of 6.5.4 is around the corner probably beginning June 2009 as the First Availability of it is due May

[bak]

I would try installing the latest HP-UX Quality Pack before jumping platforms; this can be downloaded from itrc.hp.com.  You should research how to set up persistent binding on your new Linux server because this will NOT be enabled by default.  Otherwise you will have to reconfigure tape drives after every reboot.  We experienced some strange "failure to fork" errors, just on our master server.  It turned out that HP had a defect in the way fork was implemented which was fixed in a recent patch.