On 9/24/2012 3:32 PM, Ralf Hildebrandt wrote: > >> I'm running 2.3.3 on CentOS 5 as a mail relay; most of my mail is >> delivered to an internal Exchange 2010 environment with two Hub >> Transport machines clustered behind Windows NLB under the same >> hostname. > > We have the same setup, but with a more recent postfix and OS. > Shouldn't matter that much. > > So I went looking for similar log entries in my old logs: > > Aug 27 11:44:28 mail postfix/smtp[16154]: 3X57Tg6GwvzCsRj: lost connection > with s-mx14-ht01.charite.de[10.32.37.105] while receiving > the initial server greeting > but then the 2nd internal MX was used (exchange.charite.de has two MX > hosts, pointing to the two hub transport servers). > > Sep 19 11:06:04 mail postfix/smtp[18775]: 3XMFY40JJzzCrvh: lost connection > with s-mx14-ht01.charite.de[10.32.37.105] while receiving > the initial server greeting > and again, the 2nd internal MX was used. > > So, it's happening here as well. Twice in 4 weeks.
Apparently Linux and Windows TCP window scaling doesn't always work reliably together. Try disabling TCP window scaling on the Linux box(en): ~$ echo 0 > /proc/sys/net/ipv4/tcp_window_scaling If that doesn't fix it toggle it back on with 1. To make it permanent add this to /etc/sysctl.conf net.ipv4.tcp_window_scaling = 0 -- Stan
