[Greylist-users] Strange Occurrences when MySQL Database Goes Away

[Ade Fewings]

Hi all,

Further to my message of a couple of weeks ago, I have further 
investigated the problems we have been having when the MySQL database 
behind relaydelay goes away.  For what it's worth, we're running on 
Solaris 9 SPARC, Sendmail 8.13.3, relaydelay 0.05 prerelease and MySQL 
4.1.10.

I have the following in relaydelay.conf:

...
$check_stale_db_handles = 1;
...
$pass_mail_when_db_unavail = 1;
...

When I deliberately kill the MySQL server, sendmail continues to connect 
to the milter ok, but the milter responds with 'tempfail' for all 
mails.  This is not what I would expect with the above configuration - 
am I correct or confused?  Additionally, once I then restart the MySQL 
server, the milter reconnects to it and then the first mail that 
sendmail handles talk fine to the milter.  However, relaydelay then dies 
completely and so sendmail accepts the messages anyway (by design). 

At the moment I am still in testing and so not normally tempfailing any 
messages at all - operating in 'learn' mode only to test scalability and 
reliability.  Thus, the tempfailing that occurs as described above is 
definitely out-of-the-ordinary.

Has anybody else any similar experience or advice?

Cheers
Ade