Date: Wed, 07 Feb 1996 13:25:37 -0800 From: Fred Gilham <gilham@csl.sri.com> To: stable@freebsd.org Subject: I can crash freebsd-stable reliably Message-ID: <199602072125.NAA01716@lily.csl.sri.com>
next in thread | raw e-mail | index | archive | help
After troubleshooting a problem that was bringing one of our servers down, I've made the following discovery: I can reliably crash our systems using FreeBSD-stable, 3 disks and an adaptec 2940 controller. The system is a pentium 100 with an OPTI chipset motherboard. It has 32 megs of memory. It has 3 drives. Two are Seagate 4gig hawk drives, the other is a Fujitsu 512-meg drive. (The in-use server that caused my investigation is configured similarly except it has an IBM drive instead of the Fujitsu.) We found that the in-use system would crash every night during backups (we use amanda to do backups). The system will also crash during the parallel fsck when it reboots and tries to fsck all the drives at one time. Usually it finishes one or more drives before it crashes so rebooting again will succeed, as the fsck usually only tries to do one drive. I can reliably crash the system by doing dumps of the two seagate drives at the same time. It takes less than five minutes to do this. I do dump 0sf 100000 /dev/null /dev/rsd1s1e and dump 0sf 100000 /dev/null /dev/rsd2s1e at the same time. The symptoms are that the activity light on the system goes hard on as well as the light on one of the disk drives. Also, the console message starts repeatedly reporting ahc0: target 2 lun 0 (sd2) timed out and ahc0: target 1 lun 0 (sd1) timed out After that the whole disk subsystem seems to go south. I'm using FreeBSD 2.1-STABLE on these systems (the sup is from a week or two ago). I went to this version because 2.1R was having similar problems with this setup. I was under the impression that the adaptec driver had had some changes that would help with this. I've tried using the scsi-select utility to change the speed of the scsi bus to 5mb/sec, but that doesn't cure the problem. This is a big problem for me. Can anyone help? I can supply more information or cooperate in other ways if needed. -Fred Gilham gilham@csl.sri.com
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199602072125.NAA01716>