tcpdump -n -v host sbanetweb.com and port 587 tcpdump: listening on
eth0,
This filter is too narrow to capture icmp traffic, for which the
concept of a port does not apply, and because the source address could
belong to some switch or router. Better would be:
tcpdump -n -v 'icmp || tcp port 587'
As Betarays said, a traffic capture on the destination would also
provide more light. Host- and port specific filtering is ok and
probably necessary there to minimize the noise.
Could a) rebooting the switch and b) rebooting the router at the NAS
location worth the effort.
In the windows world that seems to often be a solution... but I think
it is less effort in the long run to identify the root cause and make
specific changes in response. But the effort/reward trade-off is an
individual (or organisational :-) calculation.
MCQ (as does all the servers mentioned) have a internal IP (192 rnage)
through a NAT server to different public IP’s. They also have VPN
static ip’s.
Multiple paths and multiple addresses are of course extra complexity,
which I don't think we on this list are going to easily understand or
fix in a few words.
I changed the relay address from my domain/public address to my VPN ip
address and mail was sent.
...
Why it's going two different ways (consistently over multiple attempts
is bizarre).
What does that say?
That there is clearly a network issue, and not obviously an OpenSMTPd
one. So this is probably not the right place to solve it \_(ツ)_/.
--
Mark Lawrence
