John (and others), I completely agree that ramdisk access is a lot cheaper than mysql or some other form of DB, but I'm not sure about LDAP. It's pretty fast:
/usr/bin/time ldapsearch uid=user > /dev/null 0.00user 0.00system 0:00.00elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+373minor)pagefaults 0swaps Anyway, I do want to look at doing it the ramdisk way, so I'm wondering which file it should go into, and if there are already internal methods/functions in mimedefang that will let me read my data back out without writing a parser. David, if this isn't already done, perhaps a function/method to serialize a hash to a spool file and back would be very handy and global enough in scope to be useful for many more applications than this. Perhaps you would entertain this going into a future release? Thanks, schu John Scully wrote: > generally speaking, if your working dirs are on ramdisk it is pretty > light load to write files there. We use a mysql based mail platform and > cache information to the working dir to reduce multiple DB hits. > > When in the RCPT phase we hit the DB to see if it is a valid recipient > etc. This DB dip also returns the spam settings for that user, which we > write (append) to a file in the mimedefang working dir. > > Then, if there were any valid recipients we hit DCC, greylist, > spamassassin etc, score the message, read the spam_settings file back in > and process it for each user as per their individual settings. > > There may be lower overhead solutions, but this is pretty light, and > easy to implement and debug. > > John _______________________________________________ NOTE: If there is a disclaimer or other legal boilerplate in the above message, it is NULL AND VOID. You may ignore it. Visit http://www.mimedefang.org and http://www.roaringpenguin.com MIMEDefang mailing list [email protected] http://lists.roaringpenguin.com/mailman/listinfo/mimedefang
