On Wed, Feb 09, 2005 at 09:10:39AM -0800, Kris Kennaway wrote: > On Wed, Feb 09, 2005 at 10:34:33AM -0600, Nathan Kinkade wrote: > > Does anyone know of a way to determine the %fragmentation on a mounted > > UFS2 filesystem? An entry showed up in messages yesterday stating that > > /usr has moved from time to space optimization yet the filesystem is > > only at about 25% of it's capacity. From what I can read it seems that > > the kernel might also make this switch if fragmentation becomes > > excessive. However, this is a busy production machine running Squid, so > > I can't conveniently umount /usr. > > Try dumpfs(8). > > Kris
I had already tried dumpfs, but couldn't find any information about actual filesystem fragmentation in the output. Erik's suggestion of running `# fsck -t ufs2 /usr` seemed to work, though I felt a little skittish about running it on a live filesystem. It found numerous errors and auto-answered "no" for all of them, though I never specified that it should do that. Does fsck just do this by default on a mounted filesystem? Also, I had tried running fsck manually earlier and the only difference between what I did and Erik's suggestion was the -t option, which I wouldn't think should have been necessary. Shouldn't fsck be able to determine the fs type by looking at the superblock? By the way, the fragmentation was as 5.1%. Quite high, and I'm wondering how it got that way? Squid? Thanks, Nathan
pgpZMqMZRB9qP.pgp
Description: PGP signature