|
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [Xen-users] [GPLPV] TCP-stack leaks unpaged memory pool on Windows 2008 Standard
Hi,
recently I've migrated our fileserver VM to a new hardware. Previously using
the slow Qemu-emulation (due to some issues with an older GPLPV version), I
decided to install the newest one. It worked fine, until about 3 days later:
the whole system came to an halt, because critical kernel services couldn't
allocate any more unpaged memory. After destroy/create it worked again - about
two days. So to keep it usable, I reboot the VM every other day.
I've monitored the "unpaged memory pool" of the Windows kernel. Luckily every
allocation needs an "tag" to be clearly identified, so finding the responsible
driver was quite easy: tcpip.sys (tag: "tcDN")
Now, that's a pretty basic service which usually never causes any trouble.
Strangely, it continuously allocated memory from the aforementioned pool but
never frees it, until the pool's exhausted and kernel service after service
fails until only the mouse movement keeps functioning.
So I suspected some misconfiguration of the network. As the only thing I've
changed there was using the XenNet driver and I've read something about issues
with checksum offloading, I've tried various settings. Without avail.
Finally, I've deactivated the GPLPV drivers using the "NOGPLPV" kernel argument
- since then it runs with the Qemu-devices again and the pool, or
tcp/ip-driver, behaves normally until now.
Has anyone experienced something like this? - Unfortunately I've found nothing
on the 'net regarding this issue. Only some pages which address
debugging/analyzing self-programmed drivers regarding memory leaks.
My setup is as follows:
Hardware:
Dual-Xeon E5620 Quad-Core
24 GByte RAM
LSI RAID controller with about 6TByte storage
Intel 82567LM-2 and 82574L Ethernet
Xen:
Xen 4.1.1 (vanilla, self-compiled)
Dom0:
Ubuntu 10.04LTS
Vanilla, self-compiled 3.0.4 PV-OPS Linux kernel
self-compiled DRBD 8.3.11 kernel module & utilities
all VM storage backends on DRBD, stacked upon LVM
DomU:
Windows Server 2008 Standard (32bit)
4 Cores (exclusive)
4 GByte RAM
GPLPV: gplpv_Vista2008x32_0.11.0.308.msi
1 Ethernet device, bridged with physical
I'd be happy for any suggestion. If there's any information you need not
included above, feel free to ask. I want my 90MByte/s network bandwidth back. ;)
Kind regards / Mit freundlichen Grüßen
Björn Schneider, MSc.
IT-Administration
--
tesa scribos GmbH
Sickingenstr. 65
69126 Heidelberg
Tel.: +49 6221 33507 - 47
Fax: +49 6221 33507 - 11
Email: Bjoern.Schneider@xxxxxxxx
Internet: www.tesa-scribos.de
tesa scribos GmbH, Heidelberg. Geschäftsführer: Dr. Christoph Dietrich
(Vorsitzender), Peter Kuich, Dr. Steffen Noehte; Registergericht Heidelberg,
HRB 337200.
__________________________________.
This electronic transmission is strictly confidential and intended solely for
the addressee. It may contain information which is covered by legal,
professional or other privilege. If you are not the intended addressee, you
must not disclose, copy or take any action in reliance of this transmission. If
you have received this transmission in error, please notify us and delete the
received data as soon as possible.
This footnote also confirms that this email message has been swept for the
presence of computer viruses.
_______________________________________________
Xen-users mailing list
Xen-users@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-users
|
![]() |
Lists.xenproject.org is hosted with RackSpace, monitoring our |