Am 2017-03-07 um 11:02 schrieb Jeremy Harris: > On 07/03/17 09:28, Peter Rindfuss wrote: >> We are using Exim 4.88 on Debian with a non-Debian config. We deliver to >> our mailboxes on a separate Dovecot IMAP machine using LMTP over TCP >> (SMTP transport). A few times a day we see "Connection timed out" on the >> transport. It strikes me as odd that the time between the reception of a >> message and the delivery timeout is typically around 20 seconds, whereas >> all timeout values are at their defaults, which is 5 minutes for >> connect, command and data. Any ideas on this? > > A shorter system-level timeout may be being applied, firing before > the Exim one. Have a look around /proc/sys/net/ipv4 for values > such as "tcp_syn_retries". >
Jeremy, thank you. /proc/sys/net/ipv4/tcp_syn_retries accounts for the observed timeout, indeed. I increased it from 3 to 5. Unfortunately, I got spurious timeouts again, this time after 60s instead of 20s. We had just moved Exim to a new server. On the old server with an older Suse Linux and Exim 4.75 and the same target host, we never saw any such timeout with tcp_syn_retries = 5. Peter -- ## List details at https://lists.exim.org/mailman/listinfo/exim-users ## Exim details at http://www.exim.org/ ## Please use the Wiki with this list - http://wiki.exim.org/
