Current-Users archive

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

Re: sysctl documentation



Before I found the auto options I directly set the sendsize and recvsize options to 128k.

--Michael

On Mar 13, 2008, at 13:30, David Young <dyoung%pobox.com@localhost> wrote:

On Thu, Mar 13, 2008 at 07:26:33AM -0500, Michael Graff wrote:
I've often stumbled upon a new line in my sysctl -a output, and wondered what
exactly it does.

Today's items:

net.inet.tcp.recvbuf_auto = 1
net.inet.tcp.recvbuf_inc = 16384
net.inet.tcp.recvbuf_max = 262144
net.inet.tcp.sendbuf_auto = 1
net.inet.tcp.sendbuf_inc = 8192
net.inet.tcp.sendbuf_max = 262144

*snip snip*
On a side note, our default send/receive socket values are small enough that I can only get 350 kbytes/sec using scp between a 100 GB link here at IETF and a 1 GB link back in California. Upping these to 2x their current values let me get 1 megabyte/sec, and upping them to 128k on each end made my laptop's
hard drive limit the transfer rate.

Perhaps TCP buffer auto-sizing (sendbuf_auto) is not working right?
Which sysctls did you adjust to boost the performance?

Dave

--
David Young             OJC Technologies
dyoung%ojctech.com@localhost      Urbana, IL * (217) 278-3933 ext 24



Home | Main Index | Thread Index | Old Index