WARNING: Ignoring deprecated option --unzip

Simon Jones simonmjones at gmail.com
Tue Jan 27 08:51:48 GMT 2009


2009/1/26 Jethro R Binks <jethro.binks at strath.ac.uk>:
> On Mon, 26 Jan 2009, Julian Field wrote:
>
>> On 26/1/09 18:23, Kevin Miller wrote:
>> > -----Original Message-----
>> >
>> > > its really strange that's all, the system is easily managing to
>> > > keep up now but telnet to 25 is still really slow to respond,
>> > > like 5 - 10 seconds or it'll time out completely.  I noticed my
>> > > mailscanner sql db is getting a bit fat so maybe this is causing
>> > > some problem...
>> >
>> > Getting into the game late here, so maybe it's been asked already, but
>> > are you running a caching DNS server on the box?  That may help.
>> >
>> > Too, I've seen telnet spin it's wheels when there was no reverse zone
>> > for the source (i.e., your) host.  You don't actualy need a reverse
>> > entry for your host if you don't have one, but having even a single
>> > entry in the reverse zone allows the name server to return a 'not found'
>> > almost instantly, whereas it seems to wait to timeout if no reverse zone
>> > exists...
>> >
>> Yes, that's exactly what I was going to suggest. If telnet 25 produces a
>> long delay before giving a sendmail welcome prompt, it's a *sure* sign
>> of trouble resolving DNS names, as sendmail does forward and reverse
>> lookups on your address to work out who you are before it talks to you.
>
> No ... it's a sure sign of the system producing a long delay.  You cannot
> read any more into it than that, in the generic case.  The system may be
> heavily loaded, or delays might have been artificially introduced as a
> method of shedding spambots cheaply early on.  My own systems delay for
> different numbers of seconds at several points in an SMTP transaction,
> which throws off poorly-written SMTP engines (and ill-configured ones).
>
> But I agree DNS is probably the _likeliest_ cause in this particular case
> ...
>
>
> --
> .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .
> Jethro R Binks
> Computing Officer, IT Services, University Of Strathclyde, Glasgow, UK
> --
Thanks again to everyone for taking time to offer up suggestions -
quite a lot to go through here...

The gateways are running normally again this morning, telnet to 25
nice and quick and no mail on hold.  They do have ptr's setup so its
not a reverse dns issue and I agree in that it does smell of slow dns
but as they're on the same lan as my public name servers there
shouldn't be an issue with lookups and the name servers were
responding normally despite the problems with the gateways.  I
actually have them query the secondary name server in order to reduce
load on the primary and also cache locally on each gateway - yep, I've
been in dns hell before :)

I'm tending to lean more towards the mysql db being responsible but
I'm still looking in to it.  I have a 4gb mailscanner table which is
rather fat I feel, the db server is a quad oppy with 10gb of ram so it
has plenty of horses to play with and isn't paging (yep been there
before too...) the gateways also cache mysql lookups and I've reduced
the amount of stuff it stores in past tweaking but same past
experience tells me that slow db access has the same symptoms of that
of dns trouble.  Given the dns is and has been working fine I think it
has to be something going on with db access.

I'll check some more and post my findings.

Simon


More information about the MailScanner mailing list