[ Date Index ] [ Thread Index ] [ <= Previous by date / thread ] [ Next by date / thread => ]
Simon Williams wrote: > > Of course the question is- how do I fix the root filesystem when all I > have is ssh access? Not easily, as Neil says if you have console access somehow via KVM or similar. A reboot might work, but you may find if root partition isn't writable you can't shutdown (and trying may hang ssh access). I'd kill of what can be killed, unmount the other file systems if you can, and try /sbin/reboot, assuming I had a nice person with physical access to pick up the pieces if/when it goes pear shaped. > And I'm sure that the samba log files having filenames of ever > increasing length isn't helping- for some reason smbd.log and nmbd.log > are not being handled correctly by syslog, and end up with names like > smb.log.1.1.1.1.1.1.1.1.1.1.1.1.1.1.1.1.1 (these get quite long- I have > to delete them every so often). Any ideas? Your logrotate sounds broken, but I doubt that is the issue. It ought to do 254 character plus null easily enough, and it should error when you run out, not corrupt the disk. Most likely cause is a disk fault. Being a new disk doesn't exclude this, indeed if it is bery new, it may well have used this section of the disk for the first time in anger. Possibly some sort of logical fault in the filesystem (which hopefully fsck would fix), possibly caused by repeated loss of power? A kernel bug would be a contender, but I think unlikely. Since the disk would be made read-only on a fault, fsck will show errors, as the disk wasn't cleanly unmounted. More crucial is are any of the errors the kind that wouldn't be caused by unclean shutdown. i.e. Bad block, or other serious errors. If the server is important I'd replace the disk anyway. life is too short, hard disks are cheap. Simon -- The Mailing List for the Devon & Cornwall LUG http://mailman.dclug.org.uk/listinfo/list FAQ: http://www.dcglug.org.uk/linux_adm/list-faq.html