|
|
|
|
|
|
|
|
|
|
xen-devel
xc_dmesg.py was very helpful.
I can semi-reproduce the faults at this point. Some hosts seem more
prone than others to faulting, even though they are the same hardware.
Domain-0 has crashed rarely but normally the domain running the app
crashes. The app is a big multithreaded overlay network project
program.
I turned off -NODEBUG in arch/i386/Rules
but didn't notice any change in the console output.
I see:
DOM2: __alloc_pages: 0-order allocation failed (gfp=0x20/0)
That message doesn't always reboot domain, sometimes prints just once,
and sometimes prints in long batches.
DOM1: Weird failure in hard_start_xmit
This pops up occasionally and may prevent some TCP connections. Not
sure though.
DOM3: Unable to handle kernel paging request at virtual address c3f77820
...
DOM3: Oops: 0002
The application domain has a 1GB virtual disk swap space. It is
certainly not running out of swap. As long as domain-0 doesn't crash I
can get the oops info out of the xen console.
The setup is xen-1.2 without -NODEBUG on uni-P4 cpu with hyperthreading.
The NIC is a tg3. Running two domains, each with 1GB of swap. I'm not doing
anything at all with the balloon system.
I have not tried this with 1.3 yet.
-------------------------------------------------------
This SF.Net email is sponsored by: IBM Linux Tutorials
Free Linux tutorial presented by Daniel Robbins, President and CEO of
GenToo technologies. Learn everything from fundamentals to system
administration.http://ads.osdn.com/?ad_id=1470&alloc_id=3638&op=click
_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxxxx
https://lists.sourceforge.net/lists/listinfo/xen-devel
|
<Prev in Thread] |
Current Thread |
[Next in Thread>
|
- [Xen-devel] memory fault,
David Becker <=
|
|
|
|
|