Forum Discussion

DPO's avatar
DPO
Level 6
11 years ago

Linux client with millions of files (nfs mount points)

We have a linux client (physical server)with millions of files on its nfs mount points. It is running with NetBackup 7.5. We are trying to take the backup of these nfs mount points with NFS enabled on policy setting. Increased the client read time out to 1200. Backup is running too slow and is failing with file read failed after running for some time. The total mount points backup size is around 6 TB(with millions of files in each mount point).

 

Do we have a better solution for backing this up ?

  • You can ramp the timout right up and try again, using the accelerator option to help on subsequent runs once you manage to get that first run out of the way.

    You could also split it up into more datastreams based on its sub directories so that they all get running at the same time to reduce the overall time of the backup - again you could use accelerator with this method.

    Hope this helps

  • You can ramp the timout right up and try again, using the accelerator option to help on subsequent runs once you manage to get that first run out of the way.

    You could also split it up into more datastreams based on its sub directories so that they all get running at the same time to reduce the overall time of the backup - again you could use accelerator with this method.

    Hope this helps

  • Hi Mark,

     

    Thanks for the response. I will try as you said.. However I don't want to compromise on restore time as well.  Accelerator is already enabled. Now I'm taking the first time backup for each mount point one after the other.

  • Is there a reason you are backing up using NFS versus directly with the NetBackup client? Are these file systems from another system that are NFS mounted on the Linux client, or they are local mount points, but shared via NFS (which again I would ask: why backing up with NFS, unless you can't install a NetBackup client on this system)...

  • I'm not sure if I understand your response correctly or not.

     

    Our linux client has few NFS mount points mounted locally. We are taking the backup of this client (on which netbackup client is installed).

     

  • If I understand correctly you have locally mounted NFS filesystems from a NFS Server and are trying to back these up from the NFS Client.

    It would be better to backup from the NFS Server than the NFS client and not follow NFS mounts on each client.

    With those file numbers filesystem performance is going to be affected already and adding NFS into the mix will make it slower again.

    Check the difference between running a find . command on the NFS client and on the nfs server in one of the mount points.

    This article might help:

    http://www.symantec.com/business/support/index?page=content&id=TECH11479

  • I agree with above advice.

    Rather install NBU client on the source and backup directly from there. Unless it is a NAS filer?