Date: Wed, 27 Aug 1997 16:20:00 -0400 (EDT) From: Thomas David Rivers <ponds!rivers@dg-rtp.dg.com> To: ponds!netspace.net.au!mheath, ponds!sdf.com!tom Cc: ponds!FreeBSD.ORG!freebsd-hackers Subject: Re: Status of the following bugs Message-ID: <199708272020.QAA06960@lakes.dignus.com>
next in thread | raw e-mail | index | archive | help
Tom Samplonius wrote: > > On Wed, 27 Aug 1997, Mark Heath wrote: > > > On Tue, 26 Aug 1997, Tom Samplonius wrote: > > > > > The mangled entry error generally means that the filesystem is > > > corrupted. You need to reboot and run fsck until it comes up clean. > > > > This has been done a number of times. And the filesystem has even re > > newfs'd a couple of times. > > > > Aparently its is caused by a vnode race condition. I dont know much else > > about the bug. > > Possibly, but I haven't seen this panic in a long time. At one point, > fsck would not fix some file system problems, so every time a certain > directory was touched, the server would panic. fsck was fixed though. > > Strange thing, is that you are getting three different panics. > If you have a second - go back in the mail logs for freebsd hackers - I describe a scenario that possibly explains a couple of different panic's I get on a daily basis with my news server. Basically; it appears to be a vnode or other race condition that causes bad data to be written to the disk. The resulting panics include: ffs_alloccg: map corrupted panic: bad dir panic: ifree: freeing free inode I get one of these about every-other-day now... (look for "daily panics" in the mail logs.) When it happens, fsck usually has to be run a couple of times to fix the file system - and it happens after a newfs... I've demonstrated it on an 8-meg 386dx SCSI and IDE and a 24-meg 486dx-66 IDE box. It doesn't seem to be directly hardware related... - Dave R. -
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199708272020.QAA06960>