Hi,
Sorry for not digging the list archives before asking stupid questions :D
I have successfully built dbmail from svn trunk and populated the headers
table, however the situation did not improve much. Memory usage is still
quite high, peak at 200MB or something, CPU usage was above 80%, and most
importantly, I cannot complete the whole download process....the dbmail
process would apparantly halt at some point.
The queries generated by svn trunk as seen in mysql query log are:
SELECT physmessage_id,headername,headervalue FROM dbmail_headervalue v JOIN
dbmail_headername n ON v.headername_id=n.id WHERE physmessage_id IN (1) AND
lower(headername) IN
('references','x-ref','x-priority','x-msmail-priority','x-msoesrec','newsgroups')
SELECT physmessage_id FROM dbmail_messages WHERE message_idnr = '2'
SELECT messageblk FROM dbmail_messageblks WHERE physmessage_id = '2' ORDER
BY messageblk_idnr
SELECT physmessage_id FROM dbmail_messages WHERE message_idnr = '2'
and the query (single query?) generated by 2.0 is just
SELECT block.messageblk FROM dbmail_messageblks block, dbmail_messages msg
WHERE block.physmessage_id = msg.physmessage_id AND msg.message_idnr = '368'
ORDER BY block.messageblk_idnr
FYI, the memory and CPU consumption in 2.0 are really 'sane': cpu ~40% and
memory ~20MB.
What other information should I gather in order to understand the problem
more thoroughly? I do not have the time (yet) to look into the source code,
please forgive and correct me in case the information concerned is readily
available.....
----- Original Message -----
From: "Paul Stevens" <[EMAIL PROTECTED]>
To: "DBMAIL Developers Mailinglist" <[email protected]>
Sent: Sunday, January 22, 2006 6:06 AM
Subject: Re: [Dbmail-dev] dbmail-imapd memory usage
* mc ([EMAIL PROTECTED]) [191216 15:27]:
Perhaps I was not clear about my point; what I mean is storing each
header
name:value pair in its very own row, something like the one mentioned in
the wiki. (http://www.dbmail.org/dokuwiki/doku.php?id=headercache).
Certainly I know the header and body are stored separately as there is a
is_header flag in the table :-)
You were clear, but misunderstand: header names and values *are* stored
seperately in 2.1. Without the new tables 2.1.3 will not work properly.
And you need to fill the new tables for your existing store using
dbmail-util. As mentioned in the wiki, the new headercache setup is
fully implemented and marked as /done/.
By the way, how can I build from svn trunk? I have tried aclocal ->
automake -> autoconf -> autoreconf -fi -> ./configure, but configure
reports this:
buildtools/ltconfig: buildtools/ltconfig: No such file or directory
What should I do to build from svn sources?
I use autoreconf -i on a daily basis. Could be there's some kind of
fallout from the changes in Makefile.am lately, but that does seem
unlikely. Make sure your using autoconf-1.9 though.
--
________________________________________________________________
Paul Stevens mailto:[EMAIL PROTECTED]
NET FACILITIES GROUP PGP: finger [EMAIL PROTECTED]
The Netherlands________________________________http://www.nfg.nl
_______________________________________________
Dbmail-dev mailing list
[email protected]
http://twister.fastxs.net/mailman/listinfo/dbmail-dev
