Managed to capture and strace a snippet

sendto(40<socket:[11473780]>, "<22>Aug 19 05:38:37 dovecot: imap-login: Login: user=<matt.bryant@XXXXXX>, method=PLAIN, rip=127.0.0.1, lip=127.0.0.1, mpid=30681, secured, session=<2b5XajStxtp/AAAB>", 183, MSG_NOSIGNAL, NULL, 0) = -1 ECONNREFUSED (Connection refused) > No DWARF information found close(40<socket:[11473780]>) = 0 > No DWARF information found socket(AF_UNIX, SOCK_DGRAM|SOCK_CLOEXEC, 0) = 40<socket:[11479193]> > No DWARF information found connect(40<socket:[11479193]>, {sa_family=AF_UNIX, sun_path="/dev/log"}, 110) = -1 ECONNREFUSED (Connection refused) > No DWARF information found close(40<socket:[11479193]>) = 0

and the usual connection

u_dgr UNCONN 0 0 * 11469151 * 11469130 users:(("log",pid=28694,fd=38))

seems to just drop out of existance, the above connection on 2.2 though does reconnect ...

Has anyone else experienced this issue ?? Not sure if dovecot or centos issue tbh .... ? Anyone have any ideas ???

rgds

Matt

On 19/8/20 5:23 pm, Matt Bryant wrote:

All,

Running dovecot 2.3.10 on a CentOS 7.x and logging via syslog seems to stop when i restart rsyslog process. Reload doesnt get it working again only a dovecot restart. Tested this with dovecot 2.2.36 and this did not exhibit the same behaviour.

systemctl restart dovecot rsyslog ----- this gets a working state

systemctl restart rsyslog ------------- this restarts syslog other processes like exim continue to log but dovecot stops

systemctl reload dovecot -------------- this has no impact logging still does not work

systemctl restart dovecot ------------- this fixes the issue.

Are there any changes to the log process between 2.2 and 2.3 that might explain this ?

rgds

Matt