Can someone explain "entries sent to bpdbm" to me and why its crippling my Netapp NDMP?

[cayde-6]

Hi,

Recently we migrated from Isilon to Netapp.  Until now, things were running well, however we are now experiencing cripplingly slow speeds on our backup jobs, and many, many error 99 failures.  One thing I have noted is that the NBU jobs ends up with massive amounts of entries to the effect of "15000 entries sent to bpdbm" - this doesn't appear to correlate to the number of files, so what is it referring to?  And is this the File History element I have been reading about?  At the moment it is taking so long to enumerate this part of the job that the old jobs on the Isilon would have been finished before this part is completed, and i see the speeds of my jobs only improve when its done?

I have tried googling but honestly cant find an answer to this bit!

Thanks

[davidmoline]

Hi @cayde-6 

The number is referencing the number of files catalogued by the backup. Have a look at this article in particular ppoint 3 which increases the number of "files" sent and reduces the load. https://www.veritas.com/support/en_US/article.100004901

It may help with the backup. Given you are now on NetApp are you also enabling incremental forever backups and/or accelerator backups - which once the first backup completes (which may be painful) should sugnificantly reduce the time for subsequent backups.

Cheers
David

[cayde-6]

Thanks for the reply - just so I am clear then, is the slow nature of it enumerating how many files to back up an issue on the filer side, or the server side, or a connection between them?  From what I can see the server resources are barely being touched, so I cant really see where the slowdown for this is occuring.

I have added the file you mentioned and am re-running some jobs now to see if a difference is made.

[davidmoline]

Hi @cayde-6

Yes the slowness is the NetApp enumeration of the directory tree and is filer side. 

One way to help may be to split the NAS backup into smaller chunks (it sounds like you may already be doning this to some extent) to reduce the number of files backed up by each policy (it is usually the number of files that is the problem rather than the total size that is the cuase of slow backups).

David