Re: [Sqlgrey-users] lost connection event

[Lionel B. <lio...@bo...>]

The Mon, 15 Mar 2010 15:20:00 +0100,
Bastien LAMOUREUX <ba...@yo...> wrote:

> Hi, using SQLGrey Version: 1.6.8-1  since a 10 days on  our
> production setup, receiving around  600 000 message per day. 2
> servers hosting sqlgrey and 1 backend postgreSQL DB for sqlgrey. It
> does work really great, thanks for this great tool ;)
> 
> 
> Only little problem, I wanna solve is that really rarely (most of
> time, 1 or 2 times a day, sometimes 0), SQLGrey does encounter a
> concatenation problem and does not succeed to enter some data in DB.
> This does trigger a lost connection and reconnection event, therefore
> sending me an email to tell me that connection to database was lost
> and a second one one second later to tell me that connection is back.
> 
> I identified 2 mails which triggered this kind of problem . Any idea
> of what was so special with theses 2 emails? Or a way to tell SQLGrey
> to wait for 3 or 4 lost connections before sending alert email?
> (false positive are a bit annoying)


What probably happened is that the origin server contacted both MX at
nearly the same time (spam servers are known to do this). The first MX
managed to recognize the email and remove it from the connect table
before the second MX tried to fetch the entry in the same connect
table. This case is hard to detect and we don't have any workable
solution in the code yet.

Lionel