Date: Tue, 11 Jan 2011 08:21:50 -0600 From: Steve Price <steve@freebsd.org> To: Miroslav Lachman <000.fbsd@quip.cz> Cc: freebsd-geom@freebsd.org Subject: Re: gmirror panic on RELENG_8 Message-ID: <AANLkTinfGP%2B13Kt83s8JL%2B5Nou=Nk7nG2KS5mvG0abZO@mail.gmail.com> In-Reply-To: <4D2C546D.8010604@quip.cz> References: <AANLkTik%2B4=UMBob=Sm9tC4LhevkPCt833Fai%2BZkUSO5B@mail.gmail.com> <4D2C546D.8010604@quip.cz>
next in thread | previous in thread | raw e-mail | index | archive | help
On Tue, Jan 11, 2011 at 7:00 AM, Miroslav Lachman <000.fbsd@quip.cz> wrote: > > There must be some odd problem in your case. I am running 8.2-RC1 with > gmirror on production machine without any panic. The system was upgraded > from 7.2. I have the exact same hardware setup running in production with 8.0. This only happens when there are 3 drives (only have 3 hot swap bays so can't try more) in the box. I've /dev/zero'd all 3 drives before doing the install to make sure the contents of the drives wasn't an issue. If I have only drive, no problem. 2 drives, no problem. The mere presence of the 3rd drive on the bus and it panics. With 2 drives I can add the 3rd drive after boot like this and it works fine. # camcontrol rescan 0:2:0 # sed -i.bak -e's,#/dev/da2s1d,/dev/da2s1d,g' /etc/fstab # mount /b The really troubling part is that I can't get a coredump. da0s1b is an 8GB swap partition. # grep dump /etc/rc.conf dumpdev="/dev/da0s1b" dumpdir="/var/crash" # df -k | grep var /dev/da0s1e 81122126 262 7472094 0% /var # sysctl hw.realmem hw.realmem: 4227690496 # ls -ld /var/crash drwxr-x--- 2 root wheel 512 Dec 23 15:00 /var/crash Unless I'm missing something var has nearly twice the amount of space to hold the entire contents of physical memory but on panic it says: Cannot dump. Decive not defined or unavailable. /dev/zero'ing the whole drive is way overkill and takes forever but even when I just zero out the last sector on each disk (which is all that glabel should be caring about) it still panics when the 3rd disk is present. Doesn't matter the order of the disks either. Pick any 2 disks and it works. Use 3 in any order and it panics. Do you see anything wrong with the way I'm trying to get a coredump? If I could get that, I could get a better idea of where/how this is happening. Thanks. -steve
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?AANLkTinfGP%2B13Kt83s8JL%2B5Nou=Nk7nG2KS5mvG0abZO>