Showing results for 
Search instead for 
Did you mean: 

buffers ????

Level 6
Partner Accredited

Many times have seen under job details

"bpbkar waited xxxx times for empty buffer, delayed xxxx times"

Though there is tuning done on NBU server

Does is co-relate to it?

how can we eliminate it or reduce the time

By increasing BUFFER size?

Connect Time out?

Read Time out?

Any idea




Level 6

I used to rely on this technote:

to check what those "waited for" (need to be careful to read whether it's bptm or bpbkar) means, generally the issue here cannot be resolved by changing buffer size and # of buffers.

But I recently also found another one below which describes what happen to a SAN media server, and get into more details of calculating the delay, really worth a look!

Level 6
Employee Accredited

No, this line is perfaectly normal, and it depends on what the values are.

"bpbkar waited xxxx times for empty buffer, delayed xxxx times"

It is the 'delayed' number that is important.

This line is relating to the 'reader' process that fills the bptm memory buffer - depending on the config, it could be bpbkar or bptm.

The bptm log will also contain a similar line ...

... waited xxx times for full buffer, delayed yyy times.

What can be down depends on how high the numbers are, and which of the two times (for a given backup for example) it the highest.

Generally for example.

If the lines show bptm "waited xxx times for full buffer, delayed 890000 times." then changing the buffer size, or increasing the number of buffers  'probably' will have no effect, as the issue is more likely to be related to the speed the data is getting to the media server , due to poor read performance or network.

There is no magic fix that says cahngeing buffer sizes will improved speeds, they might improve speeds, it depends on the reason for the prro performance.



Partner    VIP    Accredited Certified

These messages used to appear  in bpbkar and bptm logs to give you a clue if you have any performance related issues. Starting with version 7.1 they are a part of Job Details and you don't have to look through tones of logs just to find them. Now, after a quick glance you can make a guess if you faced with some perfomance issue. If you know how to troubleshoot it you can fix it. Otherwise you have to ask your colleagues how to deal with it or attend some nice Symantec courses :)