Calendar based scheduling jobs dont run...quite often.

[jim_dalton]

Ive got 6.5.2A on my new server, Sol10 and mostly things are working. At least I thought they were, but on closer inspection, my operations staff are reporting that some backups havent run. There seems to be no common thread, it looked like it was purely sap agent backups but in fact there are others. They simply dont get fired off.

I understand this is the responsibility of nbpem. Anyone else seen the same?

 Damned annoying. Note this is calendar based so for the most part its Monday-Friday and a specific time window on each day. Always worked reliably on 5.1 and earlier.  Thanks for your time,Tim

[jim_dalton]

A bit more digging and no solution, but its interesting...nbpemreq -predict <date>

shows that the pem doesnt even see some of my policies! Remarkable given that I can edit them I can see the window and the date...Ive tried -updatepolicies too, makes no difference.

[jim_dalton]

Been playing with nbpemreq -predict...very interesting. Its viewpoint can vary and it doesnt match whats in my policies...this is worrying...seems like this might have something to do with aliases (some backups use a service address which is an alias of a host address). I feel a call to symantec is required.

[Andy_Welburn]

Tim

Have you tried running the nbpemreq with different options?

e.g. nbpemreq -predict_all -date <date> -client_filter <client list>

- just to see if you can tie it down by client?

(The 'predict_all' being the same as 'predict' except output includes a due time.)

Edit:

Altho' I do note that they say:

"Note that the output from nbpemreq may change from release to release.
Because of this, Symantec does not recommend using nbpemreq in scripts"

so who knows what to expect !!

Message Edited by Andy Welburn on 08-21-2008 04:24 PM

[jim_dalton]

Andy,

predict_all is fun...

# /usr/openv/netbackup/bin/admincmd/nbpemreq -predict_all -date 08/31/2008 05:00:00 -client_filter wunboris
Predicted worklist including all jobs due to run between now and Sun Aug 31 05:00:00 2008
rl client               policy               schedule             type       due time
-- -------------------- -------------------- -------------------- ---------- ------------------
00 wunboris             Solaris_OS_Files     Daily_Inc            4          Fri Aug 22 00:00:00 2008
04 sap-bid              SAP_and_ORA_Files_BID Weekly_Full          0          Sat Aug 23 00:00:00 2008
00 sap-bid              Daily_SAP_Online_BID SAP_Daily            0          Mon Aug 25 18:00:00 2008

so it can make some sense of what its supposed to do (note the alias appears sometimes in client column). Trouble is as the window is 31st aug, there ought to be an incr. per day for all policies, a full for both Weekly_SAP_Online (which doesnt appear at all) and Solaris on the weekend.Unless its supposed to just illustrate the next backup only. And the next Daily_SAP_Online is supposed to be 18:00 tonight. Has there been a re-arrangement of the logic for scheduling?

[Andy_Welburn]

---

tim walton wrote:

Unless its supposed to just illustrate the next backup only.

---

That's the feeling I get.

---

Weekly_SAP_Online (which doesnt appear at all)

---

If that's a seprate policy I can't understand why it's not being reported.

Have you tried bumping NB?

------

Also, possible similar issues with 6.0 (we had something like this on 6.5.1) have been documented but not sure if they relate to your scenario but may be worth a browse (?):

An issue identified with the resuming of failed jobs using the "Checkpoint Restart" option can lead ...

An issue within the NetBackup Policy Execution Manager (nbpem) has been identified in NetBackup 6.0 ...

The only thing is I'm not sure the '-tables screen' options are available for nbpemreq in 6.5.2.

Maybe a call is in order? 

Message Edited by Andy Welburn on 08-21-2008 04:51 PM

[jim_dalton]

A call is already with Symantec. Tried logging over the phone...gave up after 10 mins of elevator music so had to do it over the web.

Additionally I tried to create a new policy via wizard (ok) then edit it to be calendar-based ( not ok) :gave error 238, db contains erronous or conflicting values. Now stopping all my daemons and restarting.

[jim_dalton]

...and a daemon restart hasnt improved things - it still is self-inconsistent. Ho hum...

[Omar_Villa]

I though this was fix on 6.5.2a but looks is not ask symantec for the nbpem binary who fix the schedules problem with the application policies has Oracle & Lotus, this will do it.

regards.

[J_H_Is_gone]

see if this thread is any help

https://forums.symantec.com/syment/board/message?board.id=21&thread.id=46184

[jim_dalton]

Interesting comments from all...no response from Symantec support as yet.

My issue isnt related solely to Sap and Ora agent backups (Omar) , and it gets weirder...Ive got another backup that runs at 15:00, calendar based. Two days ago it did actually run, starting at 15:10, ending at 17:42. Yesterday it decided to fire off at 17:17, two and a quarter hours after the start of the window. Nothing else was running. Its like its ignoring the window, and also the calendar, some of the time.

Just to confuse the issue even more, I used the wizard to create a new policy, very basic flat file single solaris host. It created one frequency based, then I editted it to be calendar based. And it refused to accept my changes, throws error 238: 'the database contains conflicting or erroneous entries'. No other clues as to what is conflicting, I guess its in some logfile