Re: xennet0: rx no buf under NetBSD 6.1.5 domU

[mlelstv%serpens.de@localhost (Michael van Elst)]

mbowie%rocketspace.com@localhost (Mike Bowie) writes:

>Good afternoon folks,

>I've setup a domU to process NetFlow data using wip/pmacct, and have 
>begun running into periods where the domU kernel starts complaining with 
>a dmesg stream of:
>xennet0: rx no buf
>xennet0: rx no buf
>xennet0: rx no buf
>xennet0: rx no buf
>xennet0: rx no buf

That's a shortage in the "xnfrx" pool. The Dom0 is sending incoming
packets faster to the DomU than these are processed by the DomU kernel.


A common hardware driver usually allocates mbufs (mbuf "clusters" suitable
for 1500 byte packets) for its receive ring buffer. Any mbuf received and
passed to the TCP/IP stack is replaced by a newly allocated mbuf. If there
is a mbuf shortage, the packet is dropped.

For some reason, the xennet driver uses a private memory pool for the buffers
and constructs mbufs to pass to the TCP/IP stack. But the situation is
pretty much the same, when there is a buffer shortage, packets are dropped.

-- 
-- 
                                Michael van Elst
Internet: mlelstv%serpens.de@localhost
                                "A potential Snark may lurk in every tree."