Re: ixgbe input errors at high data rates

2010-03-13 Thread rihad
rihad wrote: Turning LRO off (ifconfig x.x.x.x -lro) instantly stopped the "Fragmentation failed" errors and solved the problem. Thank you. Alas, although turning LRO off, and raising dev.ix.0.rx_processing_limit from 100 to 2048 helped much, tweaking isn't over just yet. According to www.s

Re: ixgbe input errors at high data rates

2010-03-12 Thread rihad
Turning LRO off (ifconfig x.x.x.x -lro) instantly stopped the "Fragmentation failed" errors and solved the problem. Thank you. ___ freebsd-net@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-net To unsubscribe, send any mail t

Re: ixgbe input errors at high data rates

2010-03-12 Thread rihad
rihad wrote: Turning hw.intr_storm_threshold back to 1000 solved the problem Nope, nothing solved, it was a coincidence that the traffic stayed at the high level more than a few minutes. Now we're back to this: Turning dummynet off ("pipe tablearg", ~6000 table entries) improves things a b

Re: ixgbe input errors at high data rates

2010-03-12 Thread rihad
rihad wrote: rihad wrote: Things have gotten much worse. Please help. Turning hw.intr_storm_threshold back to 1000 solved the problem, at least when dummynet is not in use. Although the IP "fragmentation failed" errors (according to systat -ip) are still there, around 5-10% of the input

Re: ixgbe input errors at high data rates

2010-03-12 Thread rihad
rihad wrote: Things have gotten much worse. Please help. Turning hw.intr_storm_threshold back to 1000 solved the problem, at least when dummynet is not in use. I wonder who was the kind soul to write that tip in the ixgbe-1.3.3 readme file... Important system configuration changes: --

Re: ixgbe input errors at high data rates

2010-03-12 Thread rihad
Maybe turning DEVICE_POLLING on would be better? According to polling(4) ixgb supports it, but not ixgbe. ___ freebsd-net@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-net To unsubscribe, send any mail to "freebsd-net-unsubs

Re: ixgbe input errors at high data rates

2010-03-12 Thread rihad
rihad wrote: 1800-2000 "fragmentation failed" errors per refresh. Sorry, that's ~15000 "fragmentation failed" errors on output at 120K input packets. ___ freebsd-net@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-net To

Re: ixgbe input errors at high data rates

2010-03-12 Thread rihad
Jack Vogel wrote: The 1.3.3 driver is two years old, and your OS is older. I would respectfully suggest that you update to 8.0 where lots of effort was put to make 10G hardware perform up to its capabilities. OK, done source-upgrading kernel+world to 8.0-RELEASE-p2 with its stock ixgbe driver

Re: Re: ixgbe input errors at high data rates

2010-03-12 Thread beezarliu
收件人: rihad 抄送: freebsd-net@freebsd.org 主题: Re: ixgbe input errors at high data rates I have typically not had the bandwidth to do performance work on the drivers, I am the only FreeBSD engineer in the wired network division here at Intel, however for 10G I have changed the rules and have been

Re: ixgbe input errors at high data rates

2010-03-12 Thread rihad
Jack Vogel wrote: The difference between things being tweaked vs not is quite dramatic, like getting only 2 or 3 Gb versus getting 9.5 when properly set up. The multiqueue stack support in 8.0 is part of the equation, and in 7.X you wont have that. > Any other tweaks beyond using the bleeding

Re: Re: ixgbe input errors at high data rates

2010-03-11 Thread Jack Vogel
-------------- >> *发件人:* Jack Vogel >> *发送时间:* 2010-03-12 02:57:15 >> *收件人:* rihad >> *抄送:* freebsd-net@freebsd.org >> *主题:* Re: ixgbe input errors at high data rates >> I have typically not had the bandwidth to do performance work on the &g

Re: Re: ixgbe input errors at high data rates

2010-03-11 Thread Jack Vogel
gt; *发送时间:* 2010-03-12 02:57:15 > *收件人:* rihad > *抄送:* freebsd-net@freebsd.org > *主题:* Re: ixgbe input errors at high data rates > I have typically not had the bandwidth to do performance work on the > drivers, I am the > only FreeBSD engineer in the wired network division here at

Re: ixgbe input errors at high data rates

2010-03-11 Thread Jack Vogel
I have typically not had the bandwidth to do performance work on the drivers, I am the only FreeBSD engineer in the wired network division here at Intel, however for 10G I have changed the rules and have been working to make the driver perform well. 10G is quite different from 1G, in order for it

Re: ixgbe input errors at high data rates

2010-03-11 Thread rihad
Jack Vogel wrote: Similarly, I have done lots of work in two years to the ixgbe driver, I would even suggest that once you have 8 installed you get the driver from HEAD. You mean the driver in 8.0 isn't good enough and I'd need to somehow mix the driver from CURRENT with the rest of 8.0? ___

Re: ixgbe input errors at high data rates

2010-03-11 Thread rihad
Jack Vogel wrote: The 1.3.3 driver is two years old, and your OS is older. I would respectfully suggest that you update to 8.0 where lots of effort was put to make 10G hardware perform up to its capabilities. Similarly, I have done lots of work in two years to the ixgbe driver, I would even sug

Re: ixgbe input errors at high data rates

2010-03-11 Thread Jack Vogel
The 1.3.3 driver is two years old, and your OS is older. I would respectfully suggest that you update to 8.0 where lots of effort was put to make 10G hardware perform up to its capabilities. Similarly, I have done lots of work in two years to the ixgbe driver, I would even suggest that once you hav

ixgbe input errors at high data rates

2010-03-11 Thread rihad
Hi, our Intel 10 GigE cards are finally here, identified as with the driver ixgbe-1.3.3 off the CD-ROM. One card is used for input, the other for output, doing traffic limiting (dummynet) and accounting in between. At data rates of about 700-1000 mbps netstat -i shows many Input errors on ix0 a