On Wed, Mar 2, 2011 at 2:44 AM, David Burgess <apt....@gmail.com> wrote:
> the NIC is sending and receiving a total of about 530 > mbit x2 during the test. This gets worse I'm afraid. I recreated my setup, substituting a GS724T switch in for the GS108E, hoping the switch might be the bottleneck. Again, testing LAN>WAN iperf throughput was a flat 500 mbps, with about 10 mbps on the return during the push test. I then moved one test machine from the WAN to OPT1 and repeated the test. This time throughput dropped to around 200 mbps, and pfsense became totally unresonsive in the UI. As soon as the test ended, the UI quickly responded to whatever I might have clicked on during the iperf test. Similarly in an ssh session on pfsense, I could type in the shell and see the characters I typed with no observable latency, but pressing enter returned the carriage and produced no further output until iperf was halted. Even if I started top running before starting the iperf test, top did not update itself until after iperf was killed. Next I changed the mtu on pfsense and my test machines to 4078, the largest supported by pfsense. This time iperf throughput dropped to 96 mbps and pfsense was similarly unresponsive during the test. These results are troubling. I will probably have to test 1.2.3 on this hardware and hope for better results. Perhaps the Yandex drivers will turn this around? http://forum.pfsense.org/index.php/topic,33345.msg175595.html#msg175595 This is an Intel DG57JG board, FYI, with on-board 82578DC GBE using the em driver. db --------------------------------------------------------------------- To unsubscribe, e-mail: support-unsubscr...@pfsense.com For additional commands, e-mail: support-h...@pfsense.com Commercial support available - https://portal.pfsense.org