Dataskq hung due to large dirs in backups?

[Icheb]

Every few days I'm getting more and more reports from unmanaged servers that the dataskq keeps hanging, is running multiple times and so on.

Due to the fact I usually recommend just a kill -9, so in that case I can't find any logs, but since I've got the same issue on a managed dedicated server now, I decided to try something different.

It seems to me the process keeps 'hanging' on large dirs; there were 3 instances of dataskq running on this particular server, so I did a killall -USR1 as was described in another thread on this forum.
This got me:

2011:08:03-08:34:33: Dataskq USR1 signal: Currently processing: getDirFilesAndDirs(/home/webserv/imap/domain.nl/c.lastname/Maildir/cur, *tlf, *tdlf, (null)) : done
2011:08:03-08:34:33: Dataskq USR1 signal: Currently processing: getDirFilesAndDirs(/remotebackups/webserv/backup/domain.nl/email/data/imap/c.lastname/Maildir/cur, *tlf, *tdlf, (null)) : done
2011:08:03-08:34:33: Dataskq USR1 signal: Currently processing: getDirFilesAndDirs(/home/webserv/imap/domain.nl/d.lastname2/Maildir/cur, *tlf, *tdlf, (null)) : done

The /remotebackups is a mount from our backupsystem, it mounts over NFS while the backup is running, and unmounts when we've detected it has been completed.
So that indicates we've got at least 1 backup running.

The other 2, I'm not sure about. The dirs are quite large. I'd estimate 3 to 4 GB per dir.

Does anyone know what DA does with the 'getDirFilesAndDirs'?
Would it be possible to find a way around this function?

 

[mr.applesauce]

That shouldnt be a problem at all. I have directories 20gb and up. You should just backup to a local disk and then run some script that moves them to your nfs later.