netbackup barfs backing up netapp volume with > 3 million files. - Veritas Net Backup

This is a discussion on netbackup barfs backing up netapp volume with > 3 million files. - Veritas Net Backup ; I have a directory tree on a volume which is ~90Gb and contains over 3.5million files. The backup of this directory tree runs for a few hours and them eventually dies with a "connection timeout (41)" message. 1) I am ...

+ Reply to Thread
Results 1 to 3 of 3

Thread: netbackup barfs backing up netapp volume with > 3 million files.

  1. netbackup barfs backing up netapp volume with > 3 million files.


    I have a directory tree on a volume which is ~90Gb and contains over 3.5million
    files. The backup of this directory tree runs for a few hours and them eventually
    dies with a "connection timeout (41)" message.
    1) I am using NDMP with a SDLT320 drive attached directly to the netapp.
    2) All of the other volumes on this netapp can be backup up with no problem
    and complete in a few hours with 8MB/s to 16MB/s transfer rate.
    3) While backing up this volume, systat shows ~2% activity for long periods
    of time on the netapp (nothing written to tape, snoop shows no messages back
    to netbackup server for 10 to 20 minutes, and the filer has very little disk
    or cpu activity). Then in about 20 seconds 30MB to 50MB is written to tape
    and a flurry of messages is exchanged with the netbackup server.
    4) I tried turning off DAR onthe netapp by using
    "options ndmpd.offset_map.enable off". This had effect in that the backup
    no longer fails after 2 hours, but I am still seeing the behavior of only
    30MB being written every 20 minutes.
    5) I called netapp and they suggested testing dumping the volume to null
    as a test.
    This works correctly, finished in minutes, and showed the dump command was
    working correctly for this volume.

    Has anyone seen behavior like this before?
    Any suggestions on how to address this?
    Thanks
    Mike Y

  2. Re: netbackup barfs backing up netapp volume with > 3 million files.


    "Mike Yung" wrote:
    >
    >I have a directory tree on a volume which is ~90Gb and contains over 3.5million
    >files. The backup of this directory tree runs for a few hours and them

    eventually
    >dies with a "connection timeout (41)" message.
    >1) I am using NDMP with a SDLT320 drive attached directly to the netapp.
    >2) All of the other volumes on this netapp can be backup up with no problem
    >and complete in a few hours with 8MB/s to 16MB/s transfer rate.
    >3) While backing up this volume, systat shows ~2% activity for long periods
    >of time on the netapp (nothing written to tape, snoop shows no messages

    back
    >to netbackup server for 10 to 20 minutes, and the filer has very little

    disk
    >or cpu activity). Then in about 20 seconds 30MB to 50MB is written to tape
    >and a flurry of messages is exchanged with the netbackup server.
    >4) I tried turning off DAR onthe netapp by using
    >"options ndmpd.offset_map.enable off". This had effect in that the backup
    >no longer fails after 2 hours, but I am still seeing the behavior of only
    >30MB being written every 20 minutes.
    >5) I called netapp and they suggested testing dumping the volume to null
    >as a test.
    >This works correctly, finished in minutes, and showed the dump command was
    >working correctly for this volume.
    >
    >Has anyone seen behavior like this before?
    >Any suggestions on how to address this?
    >Thanks
    >Mike Y


    The filer makes a 20% snapshot of the files being backups (17gig aprox) before
    it NDMP's or Dumps it to tape, is the filer runnig out of space?


  3. Re: netbackup barfs backing up netapp volume with > 3 million files.


    "jdunn5" wrote:
    >
    >"Mike Yung" wrote:
    >>
    >>I have a directory tree on a volume which is ~90Gb and contains over 3.5million
    >>files. The backup of this directory tree runs for a few hours and them

    >eventually
    >>dies with a "connection timeout (41)" message.
    >>1) I am using NDMP with a SDLT320 drive attached directly to the netapp.
    >>2) All of the other volumes on this netapp can be backup up with no problem
    >>and complete in a few hours with 8MB/s to 16MB/s transfer rate.
    >>3) While backing up this volume, systat shows ~2% activity for long periods
    >>of time on the netapp (nothing written to tape, snoop shows no messages

    >back
    >>to netbackup server for 10 to 20 minutes, and the filer has very little

    >disk
    >>or cpu activity). Then in about 20 seconds 30MB to 50MB is written to

    tape
    >>and a flurry of messages is exchanged with the netbackup server.
    >>4) I tried turning off DAR onthe netapp by using
    >>"options ndmpd.offset_map.enable off". This had effect in that the backup
    >>no longer fails after 2 hours, but I am still seeing the behavior of only
    >>30MB being written every 20 minutes.
    >>5) I called netapp and they suggested testing dumping the volume to null
    >>as a test.
    >>This works correctly, finished in minutes, and showed the dump command

    was
    >>working correctly for this volume.
    >>
    >>Has anyone seen behavior like this before?
    >>Any suggestions on how to address this?
    >>Thanks
    >>Mike Y

    >
    >The filer makes a 20% snapshot of the files being backups (17gig aprox)

    before
    >it NDMP's or Dumps it to tape, is the filer runnig out of space?
    >


    Try using the FlashBackup client, which takes a snapshot of the filesystem
    rather than having to enumerate it. For systems with a large # of files,
    the benefits are supposed to be substantial [haven't tried it myself].


    Scott

+ Reply to Thread