Original problem at the bottom of the posting.
Thanks to Michael Salehi, Kevin Sheehan and Karl Vogel for their
suggestions. The culprit seems to have been in.named, which was
suffering from a slight misconfiguration for one domain and therefore
getting transfer requests much more frequently than it ought to have
done. Normally this would not have been a problem except that each of
these requests was coming in twice, all but simultaneously; I suspect
an occasional race condition. Having fixed this, things have been a
lot more stable.
>Solaris 2.4 box, patched up to October's recommended/Y2K patches.
>Over the last couple of weeks the server has been keeling over every
>couple of days with "out of inodes" errors on various partitions.
>/usr/ucb/df -i shows <10% inode usage on all partitions. fsck -n
>doesn't show up any significant fragmentation (1.2% was the highest
>I saw). No major disk space problems (several tens of megabytes free
>The last new system software installed was wu-ftpd-2.6.0, which
>doesn't appear to be causing the problem. Other possibilities that
>have been suggested: INN (not really likely to be enough volume to
>cause the problem; small site, small feed), listproc (has been running
>OK for several years, I can't see why it would suddenly have trouble).
>We're not running any caching software on this server.
>A reboot from the console cures the problem temporarily, but it's
>hardly an ideal solution.
Get Your Private, Free Email at http://www.hotmail.com
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:13:35 CDT