On Fri, 13 Jul 2007, Marcos Laufer wrote: > Hello, > > I want to report a problem i experienced while testing OpenBSD 4.1 . > I've installed it, increased VM_PHYSSEG_MAX to 16 > in /usr/src/sys/arch/i386/include/vmparam.h to make > it work with this particular motherboard and made a > stable release. > Installed a server with it and it's working fine as an MX for > a few months until now. > The machine was crashed, no error on the screen and the keyboard > did not respond. I rebooted , it started to fsck , and > the fsck failed on /usr. So i run fsck manually : fsck -y, but > it crashes with segmentation fault, so i can't mount or > start the server. > I read on the archives that it was a problem because of running out > of swap, but i had made a 2gb swap partition, despite of that > i added a 64mb file as swap and tried fsck again, but no luck. > This time it was easy for me to reinstall everything in a new hard disk, but > i still keep the old one because i would like to learn how to fix > this , if anyone wants me to make some tests or has > any ideas on what is going on , let me know.
Start by showing the error messgae. A segmentation fault is something different than running out of memory. If fsck segfaults, I need a proper error report. See http://www.openbsd.org/report.html If fsck runs out of memory, increasing ulimit -d might help, like: # ulimit -d unlimited # fsck ... That reminds me to cook a diff to do this automatically. With filesystem getting larger an larger, more people will run into out-of-mem situations. -Otto
