Re: journaling fs and large mailbox format

From: David Malone (dwmalone_at_maths.tcd.ie)
Date: 09/30/05

  • Next message: Eric Anderson: "sysctl variable creation" 
    Date: Fri, 30 Sep 2005 21:08:49 +0100
To: Vaclav Haisman <V.Haisman@sh.cvut.cz>

    On Fri, Sep 30, 2005 at 09:43:13PM +0200, Vaclav Haisman wrote:
    > I don't think that frag, inode and block size is the main factor that makes
    > XFS work well in many small files situations. From what I have read about
    > XFS I gather that it allocates inodes on demand, that it doesn't have fixed
    > amount of them.

    Nope - not by default on a Linux 2.6.12 kernel anyway:

            David.

    11% df -i .
    Filesystem Inodes IUsed IFree IUse% Mounted on
    /dev/sda2 96384 31 96353 1% /boot
    12% mount | fgrep /boot
    /dev/sda2 on /boot type xfs (rw)
    13% @ i=0
    14% while ( $i < 96358 )
    while? touch $i
    while? @ i++
    while? end
    touch: cannot touch `96353': No space left on device
    touch: cannot touch `96354': No space left on device
    touch: cannot touch `96355': No space left on device
    touch: cannot touch `96356': No space left on device
    touch: cannot touch `96357': No space left on device
    16% df -i /boot
    Filesystem Inodes IUsed IFree IUse% Mounted on
    /dev/sda2 96384 96384 0 100% /boot

    _______________________________________________
    freebsd-hackers@freebsd.org mailing list
    http://lists.freebsd.org/mailman/listinfo/freebsd-hackers
    To unsubscribe, send any mail to "freebsd-hackers-unsubscribe@freebsd.org"

  • Next message: Eric Anderson: "sysctl variable creation"

    Relevant Pages

    • Re: xfs [_fsr] probs in 2.6.24.0
      ... xfs file systems had "gone off line" due to some unknown error (upon reboot, ... I'd suggest putting lots of I/O load on your disk subsystem ... xfs_fsr/2119 is trying to acquire lock: ... We're holding the xfs inode iolock at this point to protect ...
      (Linux-Kernel)
    • Re: shrink_dcache_sb scalability problem.
      ... was (obvious to us XFS folk) just another XFS inode caching ... We fixed a similar problem in the inode cache a year or so back by creating ... more pointers to struct dentry would hurt. ...
      (Linux-Kernel)
    • Re: do_sync() and XFSQA test 182 failures....
      ... Test 182 writes a bunch of files, then runs sync, then ... The cause of the failure is that log recovery is replaying inode ... Let's translate this into what XFS does: ... see that we cannot *ever* do a proper sync of an XFS filesystem ...
      (Linux-Kernel)
    • Re: [opensuse] mv and cpio have new behavior. How do I get the old back?
      ... means your directory got a new inode number, which in turn means all .. ... is the code in xfs that does it. ... I should have used ext3 or reiser. ...
      (SuSE)
    • Re: [opensuse] mv and cpio have new behavior. How do I get the old back?
      ... willing to bet that you moved the directory to another partition, ... means your directory got a new inode number, which in turn means all .. ... is the code in xfs that does it. ... so it only changes the ctime. ...
      (SuSE)