[Xen-users] XCP 0.5 migraton troubles

[George Shuklin <george.shuklin@xxxxxxxxx>]

I test XCP with native SUSE 2.6.34-xen kernel. It works almost perfect,
but I do a stress-test (6 cores, 10 cpuburn copies, and high IO queue)
and do infinitive loop for migration:

while true:
do
xe vm-migrate uuid= host=host1 --live
xe vm-migrate uuid= host=host2 --live
done

about after 500 migrations I got message "VM didn't acknowledge the need
to shutdown" and on local console was leftopvers of crash.

Here those lines:

[ 3218.656130]  [<c0043bb5>] do_softirq+0x65/0x70
[ 3218.656130]  [<c0043e95>] irq_exit+0x65/0x70
[ 3218.656130]  [<c027ff57>] evtchn_do_upcall+0x1d7/0x220
[ 3218.656130]  [<c039a1c3>] hypervisor_callback+0x43/0x4b
[ 3218.656130]  [<c0003227>] 0xc0003227
[ 3218.656130]  [<c027deba>] force_evtchn_callback+0xa/0x10
[ 3218.656130]  [<c0040ce8>] do_exit+0x238/0x350
[ 3218.656130]  [<c039adb4>] oops_end+0xb4/0x110
[ 3218.656130]  [<c0018e2b>] bad_area+0x3b/0x50
[ 3218.656130]  [<c039d128>] do_page_fault+0x418/0x450
[ 3218.656130]  [<c039a308>] error_code+0x30/0x38
[ 3218.656130]  [<c0074808>] processCFI+0x78/0x5e0
[ 3218.656130]  [<c00747ca>] processCFI+0x3a/0x5e0
[ 3218.656130]  [<c0074f55>] unwind+0xc5/0xec0
[ 3218.656130]  [<c0009a74>] dump_trace_unwind+0xe4/0xf0
[ 3218.656130]  [<c0009b88>] try_stack_unwind+0x98/0x190
[ 3218.656130] DWARF2 unwinder stuck at try_stack_unwind+0x98/0x190
[ 3218.656130] 
[ 3218.656130] Leftover inexact backtrace:
[ 3218.656130] 
[ 3218.656130]  [<c039cd10>] ? do_page_fault+0x0/0x450
[ 3218.656130]  [<c00066ca>] ? ptregs_execve+0x12/0x18
[ 3218.656130] ---[ end trace 1b1d9b2ed521e659 ]---




_______________________________________________
Xen-users mailing list
Xen-users@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-users