Port-xen archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: xennet0: rx no buf



On Sat, 2 Dec 2017, Manuel Bouyer wrote:
On Thu, Nov 30, 2017 at 05:15:07PM +0000, Stephen Borrill wrote:
On Tue, 21 Feb 2017, Michael van Elst wrote:
netbsd%precedence.co.uk@localhost (Stephen Borrill) writes:

I'm finding that my NetBSD 7 web-server VMs are often dropping off
the network with "xennet0: rx no buf" errors, even under only moderate
load. When this happens, the machines usually end up unresponsive and need
to be hard-rebooted.

The code is pretty naive with buffer allocation.

That doesn't make it any less annoying. Adding RAM doesn't appear to help.

Any workarounds or mitigation besides rebooting nightly? Can I pre-emptively
check for near buffer exhaustion to allow for a reboot or ifconfig down/up?

Ticket #1379 should have helped with this. Does your kernel have this
change ?

It's an autobuild kernel from July (NetBSD 7.1_STABLE (XEN3_DOMU.201707032220Z) amd64) and ticket was pulled up 26 Mar 2017, so yes it has this change:

     $NetBSD: if_xennet_xenbus.c,v 1.63.2.2 2017/03/26 16:08:12 snj Exp $


--
Stephen



Home | Main Index | Thread Index | Old Index