Red Hat Bugzilla – Bug 479343
RH5.3 x64 RC1 reboots while installing a virtual machine
Last modified: 2009-01-16 04:32:24 EST
Created attachment 328506 [details]
messages file showing both reboots
Description of problem:
While installing an RH5.3 x64 RC1 VM the system reboots. Nothing on
the screen with any error messages. Just reboots.
Version-Release number of selected component (if applicable):
2 out of 2 so 100%
Steps to Reproduce:
1. Start installing a para-virt VM with x86_64 RC1
2. Wait some time to insure installation is well under way
3. Set up to install another VM with VMM and watch the system reboot
System reboots showing nothing on the screen
System stays up so I can finish the install
HP Blade BL685c
EVA-XL BootFromSan for Dom0. Used multipath during installation
EVA-XL additional lun split for VM1 and VM2 install - 10GB each as LVs
RC1 bits for x86_64
1st time: VM1-para was installing and at some point during the installation I started getting the second VM ready to install. I got to defining which logical volume to install to and the whole system rebooted. I never got to even start the 2nd VM install. xend.log shows 16:23:56 right before the reboot.
2nd time: VM1-para was installing and I was watching to see if I could spot where in the installation it may fail. I brought up VMM and got to type in the name of VM2 and as VM1 got to loading "xulrunner" the system rebooted. I'm guessing it will happen anywhere but I wanted to give you all the data I had. xend.log shows 16:50.11 right before the reboot.
Obviously a user running program shouldn't be able to reboot a system, hence the "urgent". That plus the release schedule.
Attached are the messages file and the xend.log file. If there is any additional data Red Hat would like from this system the way it is, let me know.
Created attachment 328508 [details]
xend.log showing both reboots
We really need either a core dump or, at the least, a stack trace from the crash to see what is going on here. Can you setup kdump or attach a serial console and get this information for us?
I'll see what I can do for you
Created attachment 328600 [details]
beginning captures the OOPS that causes the system reboot
Serial line capture with printk turned on
"Unable to handle kernel paging request at ffff8800bd029f58 RIP"
The beginning of the file has the error captured and then I let the system reboot in case you can fathom anything else about the system you need.
I had VM1-para installing and after a bit brought up the VMM gui to get VM2 queued up. VM1 got to "cyrus" and froze so I'm guessing it can fail at any spot. The VM2 install was not started, just brought up the gui and started to get things ready.
Hope this helps.... jim
If it helps to debug this issue I'd like to extend an offer for someone from the Westford office to come to Marlborough just down the road and get some hands on time.
479754 may be related and that is against RC2.
OK, yes. This BZ is the same as 479754, I'm pretty sure, based on the gnttab_transfer line in the serial output. I'm going to close it as such for the time being; if it turns out not to be the case, we can re-open it.
*** This bug has been marked as a duplicate of bug 479754 ***