MailScanner under openvz

[Noel Butler]

On Mon, 2010-04-26 at 08:46 +0100, Julian Field wrote:


> >
> > So I dropped the kids to "2" and it starts, all good as it should.
> >
> > 26227 ?        Ss     0:00 MailScanner: master waiting for children, 
> > sleeping
> > 26232 ?        S      0:02 MailScanner: waiting for messages
> > 28345 ?        S      0:02 MailScanner: waiting for messages
> >
> > Except.. (yep there's always an "except")...
> > Often, a child grabs the batch, then, the other kid grabs it, 
> > processing it twice, resulting in the recipient getting two copies of 
> > the message.
> That's a locking fault.
> >

That's what I thought but all perms seem correct.


> > and.. then, if it restarts itself as per the restart timeout value, 
> > they end up like this, trying to start a child, even though it already 
> > has restarted itself and both of its kids (and we don't ever see 26227 
> > return to master state.)
> > 16328 ?        S      0:02 MailScanner: waiting for messages
> > 18049 ?        S      0:02 MailScanner: waiting for messages
> > 26227 ?        Ss     0:00 MailScanner: starting child
> What happens if you put a message or two into mqueue.in and then 
> "MailScanner --debug" ?
> Does that produce any errors or warnings?



Using postfix, but have run debug about 20 times in past week with
single and multiple batches, and every time normal.


> And what does your maillog say about the whole process? Any give-aways 
> there?
> 


Not a thing, that's just it, it gives nothing away, all looks normal.
I have asked their host providor to see if it is exceeding any limits
but going by what Anthony suggested I look at, it doesn't appear to be
going near exceeding anything.

Cheers

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.mailscanner.info/pipermail/mailscanner/attachments/20100427/71843b29/attachment.html