Filesystem hang on 3ware 6.2 system
Michael R. Wayne
freebsd at wayne47.com
Thu Feb 1 19:24:56 UTC 2007
On Wed, Jan 31, 2007 at 11:28:16AM +0200, Kostik Belousov wrote:
> On Wed, Jan 31, 2007 at 02:24:35AM -0500, Michael R. Wayne wrote:
> > We have a nubmer of similar machines that were initiallly formated
> > with 6.2 before it was released and have subsequently been upgraded
> > to 6.2-RELEASE with no issues. So, we upgraded a 6.1 box which has
> > been running fine as long as the nightly dumps do not use -L to
> > take snapshots.
> > Once it was upgraded to 6.2, we enabled -L on the nightly dumps and
> > it hung on the first try. So, I'm suspecting that 6.1 has left
> > SOMEthing on the filesystem which is corrupt. For the moment. we
> > have once again removed -L from the nightly dumps but, other than
> > a complete fsck, is there any other suggested action?
> > 6.2-RELEASE
> > twe0: <3ware Storage Controller. Driver version 1.50.01.002> port 0x9c00-0x9c0f mem 0xfc9ffc00-0xfc9ffc0f,0xfc000000-0xfc7fffff irq 20 at device 1.0 on pci2
> > twe0: [GIANT-LOCKED]
> > twe0: 2 ports, Firmware FE8S 1.05.00.068, BIOS BE7X 1.08.00.048
> See kernel deadlock debugging chapter from the developer handbook for
> instructions on how to properly report deadlock.
I replied privately with no result, so I'll try the list again.
As I read the "kernel deadlock debugging chapter", I see that it
operates on core dumps. Once the machine enters this state, my
only recourse is to power cycle (shutdown hangs forever, processes
never die, etc). No disk writes seem to survive the reboot so
if I'm going to debug it, it has to be while the system is live
I'll re-iterate that it's tied to -L on dumps, no problems since
removing that option.
More information about the freebsd-stable