Re: Network performance question
Mike Smith([EMAIL PROTECTED])@2001.04.02 16:15:23 +: It's a reasonable assumption; it sounds like you haven't tuned the FreeBSD box very well, so it's doing a lot of disk I/O. I tried the test under FreeBSD with the NetGear card too - in addition to the 3COM. It's kinda strange, but when using the NetGear card and outputting tcpdump to /dev/null there were no problems, not even many interface errors (where as writing to a file causes the network to go down and tons of interface errors about halfway through the capture). This sounds like the NetGear card has issues with other PCI bus activity. what exactly is the mainboard hardware? in which slot is the card? i recall having had severe problems on some bx tyan board with 5 pci slots. when i used slot 1 or 5 i had dropped interrupts since they were shared with i tink the onboard scsi. using the middle 3 slots the problem was gone. linux seems to handle interrupt sharing on pci differently from feebsd. /k -- Experiments must be reproducible; they should all fail in the same way. KR433/KR11-RIPE -- http://www.webmonster.de -- ftp://ftp.webmonster.de To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-stable" in the body of the message
Re: Network performance question
In message [EMAIL PROTECTED], "Chad R. Larson" writes: As I recall, Mike Smith wrote: It's a reasonable assumption; it sounds like you haven't tuned the FreeBSD box very well, so it's doing a lot of disk I/O. I seem to recall that Linux does async disk writes out of the box. You are correct. Regards, Phone: (250)387-8437 Cy SchubertFax: (250)387-5766 Team Leader, Sun/Alpha Team Internet: [EMAIL PROTECTED] Open Systems Group, ITSD, ISTA Province of BC To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-stable" in the body of the message
RE: Network performance question
-Original Message- From: Mike Smith [mailto:[EMAIL PROTECTED]] Sent: Sunday, April 01, 2001 1:25 PM To: David W. Chapman Jr. Cc: Jason T. Luttgens; [EMAIL PROTECTED] Subject: Re: Network performance question FreeBSD kinda disappointed me. It gets ~1000 interface errors on about 514000 packets. I switched the 3COM card out for a NetGear FA311 (sis driver). After receiving ~31 packets, the network goes down (can't ping/telnet anywhere). At that point I have to ifconfig down and up the interface to get it back. You're disappointed in *FreeBSD* because of this? These are *hardware* failures you're describing here... Hmmso the Linux 2.4.3 kernel is somehow accessing the hardware as to not cause hardware failures then? Jason -- ... every activity meets with opposition, everyone who acts has his rivals and unfortunately opponents also. But not because people want to be opponents, rather because the tasks and relationships force people to take different points of view. [Dr. Fritz Todt] V I C T O R Y N O T V E N G E A N C E To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-stable" in the body of the message
Re: Network performance question
On Sun, 1 Apr 2001, Jason T. Luttgens wrote: packets/second. I have another computer that is multi-boot where I do a tcpdump to listen to the packets on the network and write them to a file (tcpdump -n -w test) ... Now maybe this method of testing is not proper, or there is something on the FreeBSD box I can tweak - but at this point, I'm inclined to think that Linux 2.4.3 handles high network loads better than FreeBSD. Can someone comment on this? Running a network card in promiscious mode is rather atypical. So I would say, yes, your testing methodology isn't really testing high network load. You are testing your NIC and NIC drivers ability to run in promiscious mode only. None of that traffic is even going to the IP/TCP layer of the OS. Thanks, Jason Tom To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-stable" in the body of the message
Re: Network performance question
On Sun, Apr 01, 2001 at 02:45:16PM -0400, Jason T. Luttgens wrote: -Original Message- From: Mike Smith [mailto:[EMAIL PROTECTED]] Sent: Sunday, April 01, 2001 1:25 PM To: David W. Chapman Jr. Cc: Jason T. Luttgens; [EMAIL PROTECTED] Subject: Re: Network performance question FreeBSD kinda disappointed me. It gets ~1000 interface errors on about 514000 packets. I switched the 3COM card out for a NetGear FA311 (sis driver). After receiving ~31 packets, the network goes down (can't ping/telnet anywhere). At that point I have to ifconfig down and up the interface to get it back. You're disappointed in *FreeBSD* because of this? These are *hardware* failures you're describing here... Hmmso the Linux 2.4.3 kernel is somehow accessing the hardware as to not cause hardware failures then? That's not it at all. Remember, FreeBSD and Linux can grab packets just as fast as they come into the interface... the processor is many times faster than the network card. This is definitely a hardware issue, packets are coming too fast to handle. I'd be willing to bet that Linux simply ignores the interface errors, rather than reporting them. I think what you're seeing is not that Linux handles networking better than FreeBSD, but instead that FreeBSD is more verbose in its error reporting. The important thing to remember here is that the card--not the OS--determines whether or not to drop packets. Even at 100 Mbps, a typical processor only has to poll the card 1/10 to 1/8 of the time in order to catch every bit coming in. I should point out that virtually every real-world networking test shows FreeBSD outperforms comparably configured Linux. -- Andrew Hesford [EMAIL PROTECTED] To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-stable" in the body of the message
RE: Network performance question
Hmmso the Linux 2.4.3 kernel is somehow accessing the hardware as to not cause hardware failures then? That's not it at all. Remember, FreeBSD and Linux can grab packets just as fast as they come into the interface... the processor is many times faster than the network card. This is definitely a hardware issue, packets are coming too fast to handle. I'd be willing to bet that Linux simply ignores the interface errors, rather than reporting them. I think what you're seeing is not that Linux handles networking better than FreeBSD, but instead that FreeBSD is more verbose in its error reporting. The important thing to remember here is that the card--not the OS--determines whether or not to drop packets. Even at 100 Mbps, a typical processor only has to poll the card 1/10 to 1/8 of the time in order to catch every bit coming in. Good point. One of the things I was using to judge performance was how big of a file the tcpdump on the listening machine recorded under each OS (and the number of packets reported). But maybe this is not the right way to do this So, what would be a good way to test the performace differences between Linux 2.2, 2.4 and FreeBSD as a device to capture 100% packets off the wire and not miss any? I should point out that virtually every real-world networking test shows FreeBSD outperforms comparably configured Linux. -- Andrew Hesford [EMAIL PROTECTED] To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-stable" in the body of the message