Re: Linux and mm session handler -- does it really handle high load?

Finally, there is a user notice this problem :) It's supposed to be fixed by my patch, but I'm holding commit. I appreciate if you could help to test the patch. 1) Download latest snapshot http://snaps.php.net/ 2) Apply attached patch. patch -p0 < session.diff in php source root. -- Yasuo Ohgaki Rick Morris wrote: > I have tried mm as a session handler three times, on three different Linux > boxes. Each one was a server under heavy load, running many concurrent > sessions, so I was hoping mm would solve some performance issues, but have > been disappointed each time. > > So far, I am 0 for 3 on this, having done _everything_ by the book. I am > running completely standard stuff: Apache (latest source), MySQL, PHP 4.0.6, > or PHP 4.1.1. On two mature kernels ( 2.2.19, and 2.4.17), using either the > RPM for mm, or compiling from source, I have come up against the same wall: > > Every time I try one of these heavy-load systems with the mm session > handler, performance gets WORSE instead of better. On the latest system > (Dual 1.1 GHz, 2 GB RAM, 4-disk RAID 0+1), I was losing sessions left and > right. The system could only handle about 100-150 concurrent users before > session keys started disappearing, and users would get logged out, On kernel > 2.2.19, over a period of minutes, performance would deteriorate until I > started getting segfaults. On each system, when I went back to "files" as a > session handler, the problem went away. > > So, is there something I am doing wrong? Is there some trick to configuring > mm and PHP, that is not documented, or is there a problem with Linux and mm > itself? Has anyone here successfully used mm as a session handler? On a > high-performance system? > > Any ideas, or should I submit this as a bug? If so, is it a bug with Linux, > with the MM library, or with PHP? > > Thanks, > > Rick Morris > >