Bug 37633
Summary: | [aic7xxx] SCSI bus hangs on tape access | ||
---|---|---|---|
Product: | [Retired] Red Hat Linux | Reporter: | Kevin Range <range006> |
Component: | kernel | Assignee: | Doug Ledford <dledford> |
Status: | CLOSED WONTFIX | QA Contact: | Brock Organ <borgan> |
Severity: | medium | Docs Contact: | |
Priority: | medium | ||
Version: | 7.1 | CC: | mrjones |
Target Milestone: | --- | ||
Target Release: | --- | ||
Hardware: | i386 | ||
OS: | Linux | ||
Whiteboard: | |||
Fixed In Version: | Doc Type: | Bug Fix | |
Doc Text: | Story Points: | --- | |
Clone Of: | Environment: | ||
Last Closed: | 2003-04-12 04:27:58 UTC | Type: | --- |
Regression: | --- | Mount Type: | --- |
Documentation: | --- | CRM: | |
Verified Versions: | Category: | --- | |
oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |
Cloudforms Team: | --- | Target Upstream Version: | |
Embargoed: |
Description
Kevin Range
2001-04-25 16:25:17 UTC
We ship an alternative aic7xxx module called aic7xxx_mod.o Could you try using that instead (look in /etc/modules.conf and add "_mod" to the appropriate place if your root filesystem isn't on SCSI) My root disk is SCSI (the Quantum). I suppose I could try to make a new initial root|ram disk with the _old module on it in place of the default aic driver. Any hints on how I could do that? I have some ideas, but maybe you know a better way. I mean _mod, not _old... Use the mkinitrd command to make a new boot initrd image after changing the name of the aic7xxx module in the /etc/modules.conf file. mkinitrd will automatically use the module named in modules.conf, so once you've changed that, all initrd images it build will use the correct module. After you've made a new initrd image (and optionally added a new stanza in the /etc/lilo.conf file so that you can keep both initrd images under two different names, enter them both in lilo, and be able to boot to either one by using different lilo config stanzas), then re-run lilo to activate the changes. I made the initrd's like you said, but now I can't reproduce the problem. Maybe it only happens under load... I don't know. Perhaps you want to close the bug report until I can figure out how to reproduce it all of the time (it happened three or four times in a row when i wasn't trying, and now when i want it to happen... argh!) New news. So I had given up reproducing this bug and went back to work. Anouther user ssh'd in and did some "heavy lifting" in /tmp (mounted on sdb1 which is on the narrow chain with the tape drive). Lo and behold, lock-up. So I rebooted with the aic7xxx_mod initrd I had made this morning. I had to do a manual fsck. Then on that reboot (also selecting aic7xxx_mod) I got a kernel panic (somthing about interupt handlers...) when init was "updating /etc/fstab". That scared me back to the default kernel. I had the user who crashed the machine login and run his job again. Everything went off without a hitch. It seems to me that the problem only manifests itself on the narrow interface after the machine has been in use for a while (a few hours) or I have xmms open (not necessarily playing anything, just open). Unfortunately these last two conditions are usually found together (if I have been using my machine for a while chances are I have xmms open). I will look into this some more. (I really don't see how xmms could have anything to do with it...) Have had similar problems with the system corrupting access to the tape drive. Have tried the old an new drivers and have compiled into the kernel instead of loadable modules. Still causing probs. Can clear fault by rebooting computer (power off). Tried 2.4.5 still no go. changed hardware. Still crook. When you backup and verify its fine. Next time you do it the verify fails around 40% of the files up to 50 -60%. Entire config was working flawlessly under version 7.0. Now have to go sort out prob with windows clients ftp through 7.1. they get out o.k. but time out once they log into ftp site. I got rid of all of my Travan drives, so I don't care and can't test this anymore anyway. |