Re: SMTP test failing, when others tests don't

[Martin Pala]

Hello Len,

the SMTP error is reported by monit protocol test after successful TCP connect 
was established - monit didn't receive expected response. The telnet and nmap 
-sT do only the TCP connect part, but do not perform the request/response phase.

To problem can be related to the following bug which is fixed in upcoming 
monit-5.2.5 (to be released very soon):

--8<--
* Fix Debian bug #614984: smtp protocol test issues both EHLO and
  HELO.  Thanks to Sergey B Kirpichev for report.
--8<--

To verify the root cause, we need network trace of the communication between 
monit and the SMTP server (for example using tcpdump).

Best regards,
Martin


On Mar 27, 2011, at 8:06 PM, Len Conrad wrote:

> Since we upgraded monit to latest version last Friday, this item:
> 
> check host exch1.server.tld-SMTP with address exch1.server.tld
> if failed port 25 protocol smtp with timeout 2 seconds 5 times within 5 
> cycles then alert
> 
> This began failing on one monit machine, so we added it to another monit 
> machine, where it fails, also.  
> 
> We don't get an alert every time because the test doesn't fail 5 times in 5 
> cycles, but monit.log has lots of:
> 
> [ Mar 27 09:59:10] error    : SMTP: error receiving data -- Resource 
> temporarily unavailable
> [ Mar 27 09:59:10] error    : 'exch1.server.tld-SMTP' failed protocol test 
> [SMTP] at INET[exch1.server.tld:25] via TCP
> 
> However, from either of the two monit machines:
> 
> telnet exch1.server.tld 25
> 
> and 
> 
> nmap a.b.c.d -p 25 -sT
> 
> ... work every time, and quickly.
> 
> Len
> 
> 
> 
> 
> 
> --
> To unsubscribe:
> http://lists.nongnu.org/mailman/listinfo/monit-general