XENHVM hang-up w/ "network_alloc_rx_buffers: m_cljget failed"
moto kawasaki
moto at kawasaki3.org
Sun Jan 6 13:19:58 UTC 2013
Hi Louis,
Thank you very much for your response.
kalessin> On Jan 6, 2013, at 1:39 AM, moto kawasaki <moto at kawasaki3.org> wrote:
kalessin> >
kalessin> > - Those VMs hang up with the following messages on console:
kalessin> >
kalessin> > network_alloc_rx_buffers: m_cjlget failed
kalessin> >
kalessin> > This message is repeatedly shown so quickly, and I cannot get
kalessin> > login prompt.
kalessin>
kalessin> I got this too (by starting too much torrents at once…)
kalessin>
kalessin> My understanding is that FreeBSD is not really dynamic concerning
kalessin> memory allocations for the network, so I bumped up the limits in
kalessin> /boot/loader.conf:
kalessin>
kalessin> kern.ipc.nmbclusters="51200"
kalessin> kern.ipc.nmbjumbop="51200"
In my case,
kern.ipc.nmbclusters: 262144 <-- set in /etc/sysctl.conf
kern.ipc.nmbjumbop: 12800 <-- OS default
so, 'nmbjumbop' seems the key point.
I will try to increase it.
kalessin> netstat -m is really useful to see what's going on.
Yes it really is.
But it is not so well-documented, isn't it ?
I'd really appreciate some documents/pointers describing it :-)
kalessin> My dom0 is Linux 3.2, but I don't think that this problem is specific to
kalessin> Xen in any way.
I agree.
Thank you very much
--
moto kawasaki <moto at kawasaki3.org>
More information about the freebsd-xen
mailing list