cancel
Showing results for 
Search instead for 
Did you mean: 

NBU 7.5.0.5 backup of Documentum app with tons of files: how to optimize?

Michele_Nicosia
Level 5

Good morning All,

   i'm using the normal MS-WINDOWS backup of NBU 7.5.0.5 (with pure disk feature, no tape) to backup a VM with Documentum application.

The backup is a normal FULL (Sat) and Differential Incremental (Sun/Fri).

What i've just realized is that the INC are backing up, almost , the same amount of data of the FULL, everyday.

JOBID PARENTJOB JOBTYPE ELAPSED END CLIENT   SCHED STUNIT TRY KBYTES FILES KBPERSEC DEDUP SCHEDTYPE
752125 752067 Backup 05:07:50 08/09/2015 12:17 vm895602 INC ME-A2_PDSTU 1 329533487 1472175 18660   Differential
751391 751358 Backup 08:37:07 08/08/2015 18:38 vm895602 FULL ME-A2_PDSTU 1 437625141 1615573 16322   Full

Honestly, since i don't support that Documentum app, i can't imagine 163K Files out of 186K were really changed during the 2 days from Full to INC.

Not possible, so, i think, theres hould be somethign i'm missing.

If that is on the Documentum, ok no way since i can't manage, but if it were on the Netbackup, i would like to understand and fix it.

 

Could you give me some insights how to troubleshoot it withint this scenario?

Media and master are Linux RH, client W2k8R2.

 

Thank you.

 

Regards,

 

 

Michele


 

1 ACCEPTED SOLUTION

Accepted Solutions

GulzarShaikhAUS
Level 6
Partner Accredited Certified

The default Netbackup mechanism to perform an incremental backup is using archive bit of a file/folder. If the archive bit is set then Netbackup wil surely go and backup that file. Can u observe if most of the files have archive bit set before your backup starts. 

There is another mechanism you can use which is timestamp based. You can read more about these options in the admnguide. You need to observe the state of the files and decide accordingly. It all depends on application whether a read of a file is also considered as alteration based on the way application works and the way OS understands it.

View solution in original post

10 REPLIES 10

GulzarShaikhAUS
Level 6
Partner Accredited Certified

The default Netbackup mechanism to perform an incremental backup is using archive bit of a file/folder. If the archive bit is set then Netbackup wil surely go and backup that file. Can u observe if most of the files have archive bit set before your backup starts. 

There is another mechanism you can use which is timestamp based. You can read more about these options in the admnguide. You need to observe the state of the files and decide accordingly. It all depends on application whether a read of a file is also considered as alteration based on the way application works and the way OS understands it.

Michele_Nicosia
Level 5

Hello!

Thank you for answering.

The status is: client use Archive bits, and lookign around i've found for many of them it is 1 even if the file was last accessed in 2013.

So, it could be a case or not, but changing it to Timestamps, for this server that seems to be the most complex, due the large amount of data to backup, would be a relief?

Indeed, how do i decide (if i have to) which is the correct policy to backup a file: archive bits or timestamps, if no any specific suggestion was given to me about that?

 

Thank you.

Regards,

Michele_Nicosia
Level 5

Hi,

 maybe this explain a bit, just because i have 7.5.0.5 and thousands files:

 

https://support.symantec.com/en_US/article.TECH222226.html

 

 

Thanks

areznik
Level 5

I had this problem with Documentum before (years ago, so my memory is fuzzy). It has some process related to setting premissions on top level folders that will cascade those permissions down to all the files underneath that folder in the tree and this resets the archive bits. Try changing to incrementals based on timestamps. 

GulzarShaikhAUS
Level 6
Partner Accredited Certified

Thanks Areznik,

Looks like it should be a very workable solution. 

Hi Michele,

Can you please use timestamp mechanisn for INCR backup and let us know the outcome.

Michele_Nicosia
Level 5

Hello,

  so, after few days of Timestamps, i can say: it works partially.

I mean, now the amount of files copied is drastically decreased, as i've been noticed by the fact non of the thousdas would have benn cahnged in the menawhile, but unfortunately, every day there is a 636 ERROR, on a differente drive, D:/ .

 

That's really strange: disk O: is the one with files, disk D: has less files, but every day i found the below error:

 

8/19/2015 6:42:38 AM - connecting
8/19/2015 6:42:43 AM - connected; connect time: 00:00:05
8/19/2015 6:42:59 AM - begin writing
8/19/2015 8:18:12 AM - Error bpbrm(pid=10722) could not write FILE ADDED message to OUTSOCK     
read from input socket failed(636)
8/19/2015 8:33:42 AM - Info ph901202(pid=10727) StorageServer=PureDisk:ph901202; Report=PDDO Stats for (ph901202): scanned: 746903 KB, CR sent: 9574 KB, CR sent over FC: 0 KB, dedup: 98.7%
8/19/2015 8:33:43 AM - Info bpbrm(pid=10722) validating image for client vm895602
8/19/2015 8:33:46 AM - Error bpbrm(pid=10722) could not write EXIT STATUS to OUTSOCK      
8/19/2015 8:33:46 AM - Info bpbkar(pid=8004) done. status: 0: the requested operation was successfully completed   

What i imagine is the 2 hours of time needed to create the list, or not sure, the fact this is always starting the last, while the O:, which is bigger and more populated, start first.

Error 636 is a black daemon, since i have another server (with backup bits) which is performing the same, with less files, but every day  error 636.

 

Thank you.

 

Michele_Nicosia
Level 5

This is the error 636 from another case.

https://www-secure.symantec.com/connect/forums/random-error-636-specific-sap-backup-client-windows-media-linux

Michele_Nicosia
Level 5

???

 

My update got lost!!

 

Again: the backup for disk O:/ , with multiple files, is going well now, no more huge amount of files backuped without reason. But, now i see everyday an error on disk D:/ (less populated) with error 636:

8/19/2015 4:19:24 AM - connecting
8/19/2015 4:19:29 AM - connected; connect time: 00:00:05
8/19/2015 4:19:31 AM - begin writing
8/19/2015 6:08:55 AM - Error bpbrm(pid=13926) could not write FILE ADDED message to OUTSOCK     
read from input socket failed(636)
8/19/2015 6:24:26 AM - Info ph901202(pid=13942) StorageServer=PureDisk:ph901202; Report=PDDO Stats for (ph901202): scanned: 731448 KB, CR sent: 18460 KB, CR sent over FC: 0 KB, dedup: 97.5%
8/19/2015 6:24:27 AM - Info bpbrm(pid=13926) validating image for client vm895602
8/19/2015 6:24:31 AM - Error bpbrm(pid=13926) could not write EXIT STATUS to OUTSOCK      
8/19/2015 6:24:31 AM - Info bpbkar(pid=10768) done. status: 0: the requested operation was successfully completed   

What i think, now, is that the media has some issue, then, becasue like you see in my previous link of another problem, error 636 is occuring on a SAP backup in that case, so , i don't think it depends on the bit archive or timestamp method.

Thank you.

areznik
Level 5

As you say, the error 636 is most likely a separate issue, i'd recommend opening a new thread for it (or use the one you already have)

Marianne
Moderator
Moderator
Partner    VIP    Accredited Certified

You stopped responding to the post that you refer to above and got locked after 6 months of inactivity.

What was the outcome of that issue?

Status 636 is normally resolved with KeepAlive adjustments.