05-05-2011 04:34 PM
We ae running NetBackup 6.5.4 and we are trying to backup a HPUX 11.11 host that has a folder with 10's of thousands of files in it. I have split the backup for the host up into 2 jobs, 1 to backup everything else except this folder, and another to only backup the folder with all the files in it. The first job runs OK, however the second is failing with the following messages.
<date> <time> - estimated 0 kbytes needed
<date> <time> - connecting
<date> <time> - connected; connect time: 0:00:00
<date> <time> - begin writing
<date> <time> - Critical bpbrm (pid=639072) from client wflprod.suncorpmetway.net: FTL - dir_list_rec malloc failed. Errno = 12: Not enough space
<date> <time> - end writing; write time: 0:02:31
allocation failed (10)
Any suggestions on how to fix would be great...thanks...
05-05-2011 05:18 PM
Found a possible solution by increasing the memory for the /usr/openv/netbackup/bin/bpbkar process. Does anyone know if this is to be changed on the client side or the NBU master host?
05-05-2011 05:27 PM
I hope you guys will share information how to fix this problem. I have the same problem with Billy_the_Kid and I really need to fix it as soon as possible.
05-05-2011 07:23 PM
This comment has come from one of the HPUX admins:-
"I think it's just hit a user ulimit restriction... but you need symantec to confirm, and whether it can be overcome by simply increasing the relevant limit"
The folder that I am trying to backup has 222000 or so files in it.
Has anyone know about this?
05-06-2011 12:25 AM
Hi ,
Have you configured Flash backup option in your environment? As this takes the backup in the block level rather than File level backup so in your environment taking the Flash backup of 10's of thousand file is faster than a normal backup
Regards,
ROHIT
05-06-2011 04:03 AM
I assume you found this TN : http://www.symantec.com/docs/TECH76719 ?
2. NetBackup builds inode info once backup is kicked off and it needs program data memory to complete. If there are too many small files ( ie: 4 millions files in this case ), it needs more program data memory used by the bpbkar process.
All of the above-mentioned processing happens on the client. This means that the client is running out of memory.
05-24-2011 09:06 AM
Hi,
We are having same issue on Solaris 10 client. Could you please check and confirm whether the solution given in tech note 76719 will the same is applicable for Solaris 10 client?
03:29:39.592 [85] <32> bpbkar read_and_sort_dir_entries: FTL - dir_list_rec malloc failed. Errno = 11: Resource temporarily unavailable