Mark Martinec wrote: > Curtis, > >> Thank you. I finally managed to find this in the docs. I can see where >> this would increase load time if the hash tables are big. currently my >> amavis processes are running anywhere from 100K to 100M each. the 100M >> number is scary. I'm thinking about running amavis on a separate box >> from the rest of the mail services. Its processor intensive (its gotta >> be to rip all that stuff apart and then reassemble it all.) and not >> very nice: /usr/bin/nice I've thought about methods of increasing its >> level of niceness, but haven't due to the complexity of the tasks at hand. > > Indeed a 100 MB virtul memory footprint is pretty much normal these days > for an amavisd+SA process. A 2 GB host can take about 20 such processes > (with some shared memory usage and the unused sections paged off), and > a 4 GB host twice as much. The limiting factor is still CPU, so the usual > host memory sizes are just about right for amavisd+SA and a dual CPU.
when I took a poke around on linux with pmap it looks like the vast majority of any amavis processes memory is shared with the master. The amount shared does decrease with process usage but is still a goodly amount, I'd guess more like 20mb/process. I'd love a good way to figure the total usage by all amavis children. ------------------------------------------------------------------------- Check out the new SourceForge.net Marketplace. It's the best place to buy or sell services for just about anything Open Source. http://sourceforge.net/services/buy/index.php _______________________________________________ AMaViS-user mailing list AMaViS-user@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/amavis-user AMaViS-FAQ:http://www.amavis.org/amavis-faq.php3 AMaViS-HowTos:http://www.amavis.org/howto/
