Date: Tue, 25 Feb 2014 13:18:13 -0500 From: Ryan Stone <rysto32@gmail.com> To: Johan Kooijman <mail@johankooijman.com> Cc: "freebsd-stable@freebsd.org" <freebsd-stable@freebsd.org> Subject: Re: Intel X540 T2 Message-ID: <CAFMmRNxy43Y9S0mc4_EW7tksS5CPF2947kVu90gwGf5LVcYgTQ@mail.gmail.com> In-Reply-To: <CAHvs-HXM=xTspd93Cscmezyh7_Z1EZ0s9BdK0fgsqASmuwKpMw@mail.gmail.com> References: <CAHvs-HXM=xTspd93Cscmezyh7_Z1EZ0s9BdK0fgsqASmuwKpMw@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
This sounds like it might be the issue where large NFS transfers over TSO exceed the maximum number of segments supported by ixgbe (or other drivers). There is a thread on it: http://lists.freebsd.org/pipermail/freebsd-net/2014-January/037694.html Workaround would include: - Disable TSO on the ixgbe port with "ifconfig ix0 -tso". There would be a performance impact on the NFS server. - Decrease rsize/wsize parameters on the ESXi client mounts to 32K or lower. There would be a performance impact on both the NFS server and the clients.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CAFMmRNxy43Y9S0mc4_EW7tksS5CPF2947kVu90gwGf5LVcYgTQ>