Date: Fri, 28 Feb 2014 17:38:29 -0500 (EST) From: Rick Macklem <rmacklem@uoguelph.ca> To: Markus Gebert <markus.gebert@hostpoint.ch> Cc: Johan Kooijman <mail@johankooijman.com>, FreeBSD Net <freebsd-net@freebsd.org>, Jack Vogel <jfvogel@gmail.com>, John Baldwin <jhb@freebsd.org> Subject: Re: Network loss Message-ID: <1581623680.15157356.1393627109696.JavaMail.root@uoguelph.ca> In-Reply-To: <76CCBE2B-D89E-40AE-9A58-8F022D70913B@hostpoint.ch>
next in thread | previous in thread | raw e-mail | index | archive | help
Markus Gebert wrote: >=20 >=20 >=20 > On 27.02.2014, at 18:02, Jack Vogel < jfvogel@gmail.com > wrote: >=20 >=20 > I would make SURE that you have enough mbuf resources of whatever > size pool > that you are > using (2, 4, 9K), and I would try the code in HEAD if you had not. >=20 > Jack >=20 > Thanks for the suggestion, but I do not think it has anything to do > with resource problems. We checked netstat -m among other things all > the time when the problem was occurring, and never saw any > indication of an mbuf shortage or anything similar. Looking at the > symptoms we experienced, especially the TCP connections that never > timed out, could that even be explained by mbuf shortage? >=20 >=20 > At the time we checked, 9.2 included the most recent driver AFAIR. > But this was 3 months ago, I=E2=80=99ll check if something was commited i= n > the meantime that could help us. >=20 >=20 > What I remember now is that we did see some error counter sysctl > within dev.ix rise during the network problem. It suggested that > something went wrong on the MAC layer when sending packages. I > disabled flow control, but that did not help. >=20 >=20 > Rick was so kind to point me to this other thread here: >=20 >=20 > http://docs.freebsd.org/cgi/getmsg.cgi?fetch=3D279182+0+current/freebsd-n= et >=20 >=20 > It=E2=80=99s about FreeBSD 10 and his links are flapping, which they didn= =E2=80=99t > in our case, so I did not read that thread carefully at first. But > now the OP has found out that his issues were caused by bad > firmware, I=E2=80=99m not sure anymore, if our problems could be related = to > his (instead of the nfs/mbuf problem). What do you think Jack? Is > there a way to tell what firmware we have, and what=E2=80=99s has been fi= xed > in a more recent firmware release? >=20 In the above post, it was mentioned that booting Linux from a live-CD resulted in some info (he referred to it as "spam") in dmesg. I'd suggest trying that if you can have the box offline for a few minutes. rick >=20 >=20 >=20 > Markus >=20 >=20
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?1581623680.15157356.1393627109696.JavaMail.root>