[Mailwatch-users] Memory Bloat

[Glenn Steen]

    [ The following text is in the "ISO-8859-1" character set. ]
    [ Your display is set for the "US-ASCII" character set.  ]
    [ Some characters may be displayed incorrectly. ]

On 19/10/05, Dennis Willson <taz at taz-mania.com> wrote:
> This has nothing to do with Logging.... It has to do with custom functions for "Is Definitely Not Spam" and "Is Definitely Spam".
> Are all custom functions only done by the logger? And then why do I see the initialization code run for all custom functions for
> each of the MailScanner children? I know its allocating the HASH tables and populating them from the database each and every time.
> Are you saying that it's overwriting the previous one each time so there's only one copy in memory? If so why not just test for the
> existence of the previous one before going through all that?
>
> Dennis
>
> Glenn Steen wrote:
> > On 19/10/05, Dennis Willson <taz at taz-mania.com> wrote:
> >
> >>Well I believe I have found something that causes MailWatch to use way more memory than it really needs. I'm just learning Perl so
> >>maybe I'm wrong here and if so someone please explain it to me...
> >>
> >>Setup: MailScanner configured to have 10 children
> >>        MailWatch configured with 50 whitelist entries and 50 blacklist entries
> >>
> >>Each MailScanner child starts its own copy of MailWatches SQLBlackWhite.pm which reads the 100 total blacklist/whitelist entries.
> >>Since there are 10 children total this means the 100 blacklist/whitelist entries are duplicated in memory 10 times making it take up
> >>1000 entiries worth of memory for only 100 actual entires. This multiples out very quickly.
> >>
> >>Provided this is true and there isn't something I'm missing, where there's really only one copy (which I doubt since watching the
> >>logs show each child invoking and creating its own copy of the list), is there a way in Perl to have "shared" memory?
> >>
> >>One other question is.. If using a proper connection pool to the database, individual queries should be plenty fast enough to keep
> >>up with an MTA processing mail. Why not just do the database queries "on the fly"? This would also get the latest updates instantly
> >>instead of waiting for MailScanner to restart its children or the SQLBlackWhite.pm to reload its HASH.
> >>
> >>Just some thoughts...
> >>
> >>Dennis
> >>
> >
> > Use "ps"... How many logging children do you see?
> > That's right, of all spawned children just one (is supposed to)
> > survive. And if (or rather when) that one dies, one of the MS children
> > will spawn a new one. Perhaps a bit confusing, but rather elegant in
> > its own way;-)
> >
> > --
> > -- Glenn
> > email: glenn < dot > steen < at > gmail < dot > com
> > work: glenn < dot > steen < at > ap1 < dot > se
> >

Brainshutdown. Sorry.

--
-- Glenn
email: glenn < dot > steen < at > gmail < dot > com
work: glenn < dot > steen < at > ap1 < dot > se

------------------------ MailScanner list ------------------------
To unsubscribe, email jiscmail at jiscmail.ac.uk with the words:
'leave mailscanner' in the body of the email.
Before posting, read the Wiki (http://wiki.mailscanner.info/) and
the archives (http://www.jiscmail.ac.uk/lists/mailscanner.html).

Support MailScanner development - buy the book off the website!