My database cannot keep up with the volume this seems to generate for me. I have modified/create my own sql to purge old entries from the db.
delete from triplet where uts between (unix_timestamp()-107700) and (unix_timestamp()-108000) and count=0
I run this every minute, and it purges around 6k entries.
I have added a new index on uts,count to also make this faster at selecing the old rows.
currently I am creating around 2million rows per day, after the whitelist/blacklisting.
I have tried innodb, and it did work very well, for a week, I think the issue is from how innodb uses indexs, and your primary index is way too large, would need to add an autoinc primary key, but your table schema seems to be hard coded.
Re: Speeding up db maintence
I have just uploaded the latest version into SVN. It uses a primary id index instead of what the combined one. Something like...
alter table triplet drop primary key;
ALTER TABLE `triplet` ADD `id` BIGINT NOT NULL AUTO_INCREMENT PRIMARY KEY FIRST;
..makes the tables useable with the latest version (1.006).
Use
svn co https://greylist.svn.sourceforge.net/svnroot/greylist greylist
cd greylist/trunk/
make -f Makefile.cvs
./configure
make
su
make install
to install
I'm just testing it on our servers here and it appears to be working fine.