cancel
Showing results for 
Search instead for 
Did you mean: 

NDMP backup buffer size interpretation

Jan_Voskar
Level 3
Hi,
We have a Solaris10 box with NBU6.0 MP5 with Remote NDMP backup (the backup data goes from NDMP host to a NetBackup media server's storage unit ). Until MP5 applied, everything was OK, but after MP5 for NDMP was applied the backups of NDMP host starts running 20KB/s ( too slow ). I found, that problem is in interpretation of file "/usr/openv/netbackup/db/config/NUMBER_DATA_BUFFERS". We had the file in place before MP5 with value of 128. And this is from bptm log after MP5 applied: 

12:17:22.595 [18543] <2> io_init: using 65536 data buffer value from 
/usr/openv/netbackup/db/config/SIZE_DATA_BUFFERS_NDMP
12:17:22.596 [18543] <2> io_init: using 8 data buffers
12:17:22.596 [18543] <2> io_init: using 128 data buffer size
As you can see, regardless of SIZE_DATA_BUFFERS_NDMP file in place with value of 65536, the size of buffer is set to 128 bytes, which is value from NUMBER_DATA_BUFFERS file !
And number of buffers is set to 8, which is default value, when there is no NUMBER_DATA_BUFFERS file !
After change of value in NUMBER_DATA_BUFFERS file from 128 to 64, the bptm log looks like this:
13:10:37.535 [20449] <2> io_init: using 65536 data buffer value from 
/usr/openv/netbackup/db/config/SIZE_DATA_BUFFERS_NDMP
13:10:37.536 [20449] <2> io_init: using 8 data buffers
13:10:37.536 [20449] <2> io_init: using 64 data buffer size

Again the same issue!
The only situation, when the speed of backups are fine again, is when there is no NUMBER_DATA_BUFFERS file at all:
13:10:37.535 [20449] <2> io_init: using 65536 data buffer value from 
/usr/openv/netbackup/db/config/SIZE_DATA_BUFFERS_NDMP
13:10:37.536 [20449] <2> io_init: using 8 data buffers
13:10:37.536 [20449] <2> io_init: using 65536 data buffer size
But it is unacceptable for me, because other clients backups would be slow, because of 8 buffers is too low number for LTO3.
It looks like clear bug of NDMP MP5 patch, but maybe not.
Does anybody meet with the same issue ?
Thank you for any hints.

Jan


4 REPLIES 4

Jan_Voskar
Level 3
Confirmed by support, that it is a bug of MP5.

hungry_jack
Not applicable
We just upgraded to MP5 for v6.0... tested everything.... and immediately noticed ndmp is as slow as ever.....
logged a problem to Veritas (Symantec)....
 
and are now looking at the data buffer parameters......

migzi
Not applicable
I'm just wondering if a fix has been provided for this problem?
 
I have been troubleshooting NDMP backup performance for a week or so, and hadn't used it previous to 6.0 mp5 so had no idea why NDMP backups were running so slow. (I have been looking elsewhere not at buffer sizes, to try and fix it).
 
I then stumbled across this thread and I have confirmed the same symptoms as above.
 
So was wondering if there was any news from Symantec?
 

Stumpr2
Level 6
I can't say this technote will help but there is some mention of buffers...
 
BUG REPORT: After applying NetBackup 6.0 MP5, NDMP backups begin failing with NetBackup Status Code 27 (child process killed by signal) or Status Code 99 (NDMP backup failure)
 
.
.
.
The number following the NDMP_MOVER_SET_RECORD corresponds to the number placed in the NUMBER_DATA_BUFFERS file (found in the /usr/openv/netbackup/db/config/ directory on a UNIX/Linux media server or in the <install_path>\veritas\netbackup\db\config directory on a Windows media server).  If the number following NDMP_MOVER_SET_RECORD_SIZE is something other than 64512, the site is affected.  

Workaround:
Depending on the environment, there are two workarounds available:
1. Remove the NUMBER_DATA_BUFFERS touch file from the media server running backups for the NDMP devices.  
2. Remove NetBackup 6.0 MP5 and reapply NetBackup 6.0 MP4.  

Please be aware option 1 can potentially introduce other problems and negatively impact NetBackup performance.  Please exercise caution while testing this workaround.  If the NUMBER_DATA_BUFFERS touch file is required for the environment, please contact Symantec Technical Support for additional assistance.  

ETA of Fix:
Symantec Corporation has acknowledged that the above-mentioned issue is present in the current version(s) of the product(s) mentioned at the end of this article. Symantec Corporation is committed to product quality and satisfied customers.

This issue is currently being considered by Symantec Corporation to be addressed in a forthcoming Maintenance Pack or version of the product.  The fix for this issue is expected to be released in the fourth quarter of 2007.  Please note that Symantec Corporation reserves the right to remove any fix from the targeted release if it does not pass quality assurance tests or introduces new risks to overall code stability.  Symantec's plans are subject to change and any action taken by you based on the above information or your reliance upon the above information is made at your own risk.  Please refer to the maintenance pack readme or contact NetBackup Enterprise Support to confirm this issue (ET1111632) was included in the maintenance pack.  

As future maintenance packs are released, please visit the following link for download and readme information:  
 http://www.symantec.com/enterprise/support/downloads.jsp?pid=15143