cancel
Showing results for 
Search instead for 
Did you mean: 

Jobs kicking in before the fingersprints loads up in the cache *dedup backup*

Spartacus81
Level 6
Partner Accredited

Friends,

enviornment: Master; 2 Media servers running NBU 7.5.0.1

I'm running into this problem where subsequent backup followed the seed is failing with error 83 media open error and 10 memory allocation error. I have server sitting on 4meg line and holds over 2TB of data at remote; done the seed; modified the pd.conf file (fp_cache_client_policy) so that fingerprints could load up in the momory and using client side deduplication it process the data and only send the changed blocks over.. but its not happening job failing straight away, i have reduced the number of streams on the client but no joy.. disabling client side dedup and use media server dedup is not an option here as per the technotes becaue it will take forever to send the data over.. any help or suggestion would highly be appreicated?

 

1 ACCEPTED SOLUTION

Accepted Solutions

Spartacus81
Level 6
Partner Accredited

unable to resolve this issue.. the last response i had from symantec was:

"I apologize for the delays, I have been swamped.  The Backline Engineer reviewed the information you sent in and a few files have very old dates, close in fact to some point in history that if files have a touch date that old you get behavior as you see.  I need to chat with him about looking at something in your environment.  He mentions it in a note but I need some clarification before I call.  I will touch base with Justin in the morning and call you afterwards."

anyway workaround: apply 7.5.0.4 and use accelerator to bcakup the host.

 

 

View solution in original post

6 REPLIES 6

V4
Level 6
Partner Accredited

Try accelerator and check if network resiliency helps

BTW there was known issue of status 83/84 on DeDupe with NBU 75  Check LBN for same

Also if it's possible to keep additional MSDP at remote site and enable AIR between MSDP only. this would ensure only optimum data to be transported between Server to server

Spartacus81
Level 6
Partner Accredited

@Captain Jack sparrow:

Thanks; guess what i used accelerate and run into an issue where large TRACK.dat file being generated and eating all space on the C: drive.. we cant move that file to any other location and symantec suggests the system with loads of changes is not a good candidate for accelerate hence i had no choice except to roll back and use traditional backup again..

any other suggestion to make this backup any faster? i have even put McAfee policy in place to exclude all backup process.. is there any software that we can use to track the changes in the last 24hrs? cant think of anything else.. am struggling to find the technical justification to give to my customer.... :s

V4
Level 6
Partner Accredited

oops

 Apologies on late highlight on this new feature

i also had similar issue . unfortunately it differs from environment to environment

Few of my customer are happy with Accelerator but some are unhappy due to same journallog which fills up entire disk .Symantec sayd change journal log is set in circular logging method of 30 MB but seems there is bug with it.. May be next release would fix up.

you may try 2nd option which i gave. Thats only way you can reduce network traffic along with offloading Multiple Client connections to central NBU server

Spartacus81
Level 6
Partner Accredited

thanks but am afraid its not an option..my customer enviornment is already complicated enough..

2 sites; two master (AIR) along with 8 media servers and 2 load balancing servers at eac site.. around 300 remote sites all are on different link..

some remote sites handles the Dedup load well and some doesnt and running MSDP dedup even though you have seeded the data will just take forever.... am thinking to go to 7.5.0.4 but am not sure if that resolves the issue am facing here...

Spartacus81
Level 6
Partner Accredited

Interesting;

why job is running full and scanning all the files its because bpbkar logs says something like this..

dos_backup::tfs_scannext: INF - detected renamed/new file:<C:\Tmp\1.txt>, forcing backup

i have modified the file and back it up, NBU clearnig the archive bit as it suppose to do on that file but its backs up all other files as well.. any idea?

Spartacus81
Level 6
Partner Accredited

unable to resolve this issue.. the last response i had from symantec was:

"I apologize for the delays, I have been swamped.  The Backline Engineer reviewed the information you sent in and a few files have very old dates, close in fact to some point in history that if files have a touch date that old you get behavior as you see.  I need to chat with him about looking at something in your environment.  He mentions it in a note but I need some clarification before I call.  I will touch base with Justin in the morning and call you afterwards."

anyway workaround: apply 7.5.0.4 and use accelerator to bcakup the host.