I understand how it's supposed to work - it spawns up to your max_processes process that will stay around for awhile. And if things are being slow for some reason, it'll spawn up to another max_processes to deal with the temporary slowdown.
My problem is that if the site gets slow, it's generally because it's having ram issues. So spawning 4 more interpreters (even if they have some copy-on-write shared memory), actually ends up hurting things. I'd prefer it in this case to either be able to 1) limit the number of 'temporary' processes that get spawned, or 2) disable this alltogether and just have it queue requests/etc.
Also, I'm still having problems with some of ruby/lsapi processes that have hanged waiting for something can't be killed without a kill -9, which lshttpd never does. So sometimes I need to go in and clean up old processes by hand.
Also, the 'general' board is still called 3.0 pre-release, perhaps it's name should be updated.
Thanks for a great product!
-Kevin
My problem is that if the site gets slow, it's generally because it's having ram issues. So spawning 4 more interpreters (even if they have some copy-on-write shared memory), actually ends up hurting things. I'd prefer it in this case to either be able to 1) limit the number of 'temporary' processes that get spawned, or 2) disable this alltogether and just have it queue requests/etc.
Also, I'm still having problems with some of ruby/lsapi processes that have hanged waiting for something can't be killed without a kill -9, which lshttpd never does. So sometimes I need to go in and clean up old processes by hand.
Also, the 'general' board is still called 3.0 pre-release, perhaps it's name should be updated.
Thanks for a great product!
-Kevin