11d woes

[Ade_Sulaiman]

hello all. i have had nothing but problems ever since upgrading to 11d. lotus agent crashed the program (there is a reg fix for this), jobs could not be saved (another reg fix), jobs take extremely long times to complete - 14hr for 200gb, a single job in 10d had to be split into 4 different jobs in 11d on the advice of tech support, jobs just stall for no visible reason and at random. the time continues to run, the job is still listed as 'running' but the byte count does not increase.
i have not had a single successful job complete since the upgrade and think i have to consider down grading to 10d.
has anyone else had similar issues or is it just me..?

[S_Cooney]

We had this problem, changed the 320 SCSI card to a 160 and it sped up like lightning!

[Dennis_Thornton]

The common consensus is to do a fresh install. Upgrading doesn't appear to be fully worked out and installing from scratch has cured a lot of problems.

[Ade_Sulaiman]

hi

this was backing up to disk. although the new quantum superloader 3 doesn't fare any better speed wise. i'll check the scsi card on the server here.

[Ade_Sulaiman]

done. re-installed a fresh on a different server with no difference. i find the backups seem to stall more on system states.

[Sandy_Sandifer]

Hi Ade,

I read through your case in our database. Here are a couple more suggestions/questions that I didn't see requested in the case:
1. What happens when you perform a backup of just system state on the servers? What happens if you run a backup job without backing the system state on any of the servers? How fast does it run then?
2. On the one job that backed up successfully, but was 17 hours slow, if you look through the job log, where are the slow points? Do they occur when it hits the system state on all the servers?
3. Are you using AOFO, and have you tried running a backup of system state without AOFO checked? Please try that and reply back with the results.

At this point, even though I think my questions above are valid and need to be taken a look at, my best advice to you since you have a case history with us, would be to call back into Support and request for this case to be reopened and escalated to an Advanced Team technician.

If you do not call back in, please let me know if you can answer the questions above. Even with those answers though, I have a feeling a more thorough investigation with debugs, sgmon logs, and other logs would need to take place.

Thanks.

[Ade_Sulaiman]

hello sandy

thanks for your response, i'll attempt to answer the questions as best i can

1) i have not yet created a separate job for sys states. if i do this will be the 4th job that will come out of my original single job is v10d. i have considered doing this but because of the time issues (for example) i just needed to get jods working.

2) on the 'successful' jobs, they always have to be cancelled at the end due to the job stalling. so the time of 17hr i gave would be the time at the time the job was cancelled because the byte rate had ceased. if i check the log before i cancel it would have stopped on a sys state - not nessesarily the same one - after i cancel the logs will only display in text. from what i can tell the slow down is on file servers backing up ordinalry pdf's, doc's, xls etc.

3) the job containing the sys states is running with aofo. the database and exchange jobs are not. to get around this will mean creating a separate job for sys states.

4) i may not have mentioned; the long backup times are occuring on backup to disk. here the byte rate can drop as low as 250mb/min. backup to tape is not much better at about 450mb/min. i have a brand new lto3 library from quantum. they say i ought to be able to achieve nearly 10 x that speed to tape. i realise a lot of factors can contribute to transfer speeds but being able to clear 300gb in 8hr to doing it in between 17 and 23 hrs (to disk) is a massive fall in performance.

5) i may not be able to do much testing with 11d now as i plan to revert back to 10d this weekend. however, i would still want to use 11d (more for exchange grt) but would also need to know that i will not face any such issues when doing so. maybe 12d..?

thanks.