Limiting the number of Superblock duplicates on newfs of huge filesystem

From: Chris Ruhnke (ruhnke_at_us.ibm.com)
Date: 09/14/05

  • Next message: sunny: "Recovering the SVM volume" 
    To: sunmanagers@sunmanagers.org
Date: Wed, 14 Sep 2005 13:20:29 -0500

    I don't have to newfs filesystems very often so I don't run into this
    problem that much.

    I have just built a 500+ GByte RAID-5 user data filesystem. I am running
    "newfs" against it and it has been running for almost an hour now. Because
    of the amount of data space available, newfs is creating a gazillion
    superblock duplicates. Now, I am quite happy to have superblock
    duplicates in the event of loss of the primary superblock. But this is a
    failure that rarely happens any more. Usually the entire filesystem gets
    corrupted beyond redemption. I don't mind giving up some space to a few
    -- even a few hundred -- superblock clones. But there comes a point when
    duplicate superblocks are just a waste of space and fragementation of the
    data area.

    Does anyone have any suggestions about how to limit the number of
    superblock duplicates that get created on a ufs filesystem. I am
    restricted by the customer to using UFS and do not have the luxury of
    looking into Veritas or any other "smart" volume/filesystem manager. I
    haven't found anything promising in the man pages.

     
    --CHRis

    Chris H. Ruhnke
    Technical Services Professional
    IBM Global Services
    Dallas, TX

    O'Toole's Law: Murphy is an optimist.
    _______________________________________________
    sunmanagers mailing list
    sunmanagers@sunmanagers.org
    http://www.sunmanagers.org/mailman/listinfo/sunmanagers

  • Next message: sunny: "Recovering the SVM volume"

    Relevant Pages

    • Re: newfs and mount vs. half-baked disks
      ... >> newfs was developed with operator intervention in mind. ... >> performance hit in updating all superblock copies at the end is small ... >affecting the filesystem itself? ... My suggestion would be to write a non-standard magic number to ...
      (freebsd-arch)
    • SUMMARY: Limiting the number of Superblock duplicates on newfs of huge filesystem
      ... I am quite happy to have superblock ... > superblock duplicates that get created on a ufs filesystem. ...
      (SunManagers)
    • Re: Is this Hard drive dead, Jim?
      ... > can give a second opinion. ... you must specify the filesystem type ... > The superblock could not be read or does not describe a correct ext2 ...
      (comp.os.linux.hardware)
    • Re: newfs and mount vs. half-baked disks
      ... > clean we just newfs and mount the clean new filesystem. ... > performance hit in updating all superblock copies at the end is small ... since the superblock for the previous file system may ... Newfs should also set the unclean flags and any related flags until it ...
      (freebsd-arch)
    • Re: Failure to Mount
      ... > fstab says that hdb has 3 primary partitions, ... Filesystem magic number: 0xEF53 ... Couldn't find ext2 superblock, trying backup blocks... ... Couldn't find valid filesystem superblock.# ./tune2fs ...
      (comp.os.linux.misc)