Re: Serious Postfix weirdness
You can lower MSS instead of lowering MTU - it's much safer option. How do you lower MSS and not lower MTU? Thanks, Petr
Re: Serious Postfix weirdness
How is it possible that the sender uses wscale 2^7 and the receiver 2^0? Is this a problem? Could Sephe give a suggestion about what is the problem here and how I can solve it? Thanks, Petr
Re: Serious Postfix weirdness
On Sat, Nov 15, 2008 at 07:21:19PM +1100, Petr Janda wrote: How is it possible that the sender uses wscale 2^7 and the receiver 2^0? Is this a problem? One side says it can do window scaling and the other can't. That is not necessarily a problem, it can happen depending on the maximum TCP window sizes as the BSD stack will not advertise window scaling if it doesn't need it (IIRC). Do you have a firewall in between? At least old PF (not sure which is in src right now) had issues with window scaling. Joerg
Re: Serious Postfix weirdness
On Sat, Nov 15, 2008 at 4:21 PM, Petr Janda [EMAIL PROTECTED] wrote: How is it possible that the sender uses wscale 2^7 and the receiver 2^0? Is this a problem? window scaling shift 0 is allowed. It just mean we support window scaling, but the shift is 0. From the dump, it looks like the other side is a Linux box. Could Sephe give a suggestion about what is the problem here and how I can solve it? As Joerg has suggested, window scaling option may be problematic with pf. You could try set net.inet.tcp.rfc1323 to 0 Best Regards, sephe -- Live Free or Die
Re: Serious Postfix weirdness
window scaling shift 0 is allowed. It just mean we support window scaling, but the shift is 0. From the dump, it looks like the other side is a Linux box. Could Sephe give a suggestion about what is the problem here and how I can solve it? As Joerg has suggested, window scaling option may be problematic with pf. You could try set net.inet.tcp.rfc1323 to 0 Here is another dump this time with rfc1323, sack and smartsack disabled. Do you notice something weird here? Thanks, Petr postfix_dump2.tgz Description: application/tgz
Re: Serious Postfix weirdness
On Sat, Nov 15, 2008 at 6:17 PM, Petr Janda [EMAIL PROTECTED] wrote: window scaling shift 0 is allowed. It just mean we support window scaling, but the shift is 0. From the dump, it looks like the other side is a Linux box. Could Sephe give a suggestion about what is the problem here and how I can solve it? As Joerg has suggested, window scaling option may be problematic with pf. You could try set net.inet.tcp.rfc1323 to 0 Here is another dump this time with rfc1323, sack and smartsack disabled. Do you notice something weird here? 17:48:34.875003 IP 203.16.214.214.41558 202.76.131.108.25: P 91:145(54) ack 218 win 49640 /* 17:48:34.915642 IP 202.76.131.108.25 203.16.214.214.41558: P 218:269(51) ack 145 win 58400 */ 17:48:34.952747 IP 203.16.214.214.41558 202.76.131.108.25: P 7445:8337(892) ack 269 win 49640 Several segments (203.16.214.214 - 202.76.131.108) are never seen by the driver. 17:48:34.952792 IP 202.76.131.108.25 203.16.214.214.41558: . ack 145 win 58400 Application seems to use 5min timeout. The question is why 203.16.214.214 did not rexmit? 17:53:34.930919 IP 202.76.131.108.25 203.16.214.214.41558: P 269:325(56) ack 145 win 58400 17:53:34.931606 IP 202.76.131.108.25 203.16.214.214.41558: F 325:325(0) ack 145 win 58400 Best Regards, sephe -- Live Free or Die
Re: Serious Postfix weirdness
On Sat, 15 Nov 2008 22:39:26 Sepherosa Ziehau wrote: Several segments (203.16.214.214 - 202.76.131.108) are never seen by the driver. 17:48:34.952792 IP 202.76.131.108.25 203.16.214.214.41558: . ack 145 win 58400 Application seems to use 5min timeout. The question is why 203.16.214.214 did not rexmit? Im puzzled by this myself. Right now I have no idea where the problem. Why would the sender not retransmit? Ive been on this for 2 days now and cant get iit over with. grr Thanks, Petr
Re: Serious Postfix weirdness
If hardware checksumming is turned on on the interface, try turning it off. Broken checksum offloading can cause packets to be dropped by routers as well as end-points. -Matt Matthew Dillon [EMAIL PROTECTED]
Re: Serious Postfix weirdness
On Sun, 16 Nov 2008 08:13:34 Matthew Dillon wrote: If hardware checksumming is turned on on the interface, try turning it off. Broken checksum offloading can cause packets to be dropped by routers as well as end-points. Is hardware checksuming the options in ifconfig: hwcsum and txcsum? Petr
Re: Serious Postfix weirdness
:On Sun, 16 Nov 2008 08:13:34 Matthew Dillon wrote: : If hardware checksumming is turned on on the interface, try turning it : off. Broken checksum offloading can cause packets to be dropped by : routers as well as end-points. : : :Is hardware checksuming the options in ifconfig: hwcsum and txcsum? : :Petr Yes. Are they on? If so then turn them off and see if that fixes the problem. -Matt Matthew Dillon [EMAIL PROTECTED]
Re: Serious Postfix weirdness
Yes. Are they on? If so then turn them off and see if that fixes the problem. I turned them off but it made no difference. On the other hand I decided to lower the MTU even more, to 800 and guess what.. some of the lost mail started getting through. Unfortunately with a setting like this, imap stops working so i had to go back to 1500. Petr
Re: Serious Postfix weirdness
On Fri, Nov 14, 2008 at 08:15:04PM +1100, Petr Janda wrote: Supposedly the problem here is that the sending machine has got a firewall in front of it thats blocking ICMP MUST FRAGMENT. Is net.inet.tcp.path_mtu_discovery=1? Joerg
Re: Serious Postfix weirdness
Is net.inet.tcp.path_mtu_discovery=1? Joerg No, it was set to 0. is it supposed to be set to 1? If so, should the default be 1? As far as documentation goes Ive read most of modern UNIX systems have it turned on by default. Cheers, Petr
Re: Serious Postfix weirdness
Joerg Sonnenberger wrote: On Sat, Nov 15, 2008 at 04:31:55AM +1100, Petr Janda wrote: Is net.inet.tcp.path_mtu_discovery=1? Joerg No, it was set to 0. is it supposed to be set to 1? If so, should the default be 1? As far as documentation goes Ive read most of modern UNIX systems have it turned on by default. It might help. Joerg Does anybody object to turning Path MTU Discovery on by default? It's been more than 15 years since it was introduced and everybody had a chance to learn about it by now. Is there a technical reason (e.g. related to where the Path MTU is stored), for having it off till now?
Re: Serious Postfix weirdness
On Fri, Nov 14, 2008 at 11:05:00PM +0200, Jordan Gordeev wrote: Is there a technical reason (e.g. related to where the Path MTU is stored), for having it off till now? Stupid network admistrators that consider all ICMP traffic evil and block it. But IMO it should be active by default. Joerg
Re: Serious Postfix weirdness
It might help. Joerg Ive had it on for like 6 hours now but i dont think it made a difference. Thanks anyway. Im welcome to more suggestions. Petr