UPDATE: RH 9 Memory Util Problem?

I should have also mentioned that post-upgrade, we are running ext3 as
well, we were running ext2 before.  

After some additional googling around, I'm not sure that what I'm
seeing is a problem.
The symptom:
"Free memory as reported by top goes down regularly, with the value
eventually stabilizing, but I can't understand why"

Results from top (after some uptime and when things are fairly quiet):

 15:14:24  up 2 days,  9:22,  2 users,  load average: 0.27, 0.16, 0.17
70 processes: 68 sleeping, 2 running, 0 zombie, 0 stopped
CPU0 states:  33.4% user   3.2% system    0.0% nice   0.0% iowait 
62.2% idle
CPU1 states:  17.4% user   1.3% system    0.0% nice   0.0% iowait 
80.1% idle
Mem:  1030304k av,  926880k used,  103424k free,       0k shrd, 
142584k buff
                    518220k actv,   33692k in_d,  209492k in_c
Swap: 1020116k av,   11720k used, 1008396k free                 
549676k cached

I think that what I'm looking at is normal behavior, as follows
(excerpt from a different list):

"... it simply sounds like cache effects.  atime updates result in
journal commits under ext3, and 
those use at least a couple of buffers at a time (one for metadata
descriptor block in the journal, 
one for the journal commit.)  Those aren't leaks  they are temporary
use of cache, and once the IO has
complete the memory can be immediately reclaimed by the kernel if it is
needed for anything else."

I believe that where we are getting into trouble is when too many http
requests come too quickly and:
- the system doesn't release the resource quickly enough (it is an
older box, dual 450Mhz CPUs),
- httpd forks additional processes to handle requests (sometimes
hitting MaxClients), and,
- we eat into swap too quickly and deeply, eventually we start to
thrash

Short-term we'll drop maxclients down, using it as a brake to avoid
taking the box down.  
Longer-term we'll look at moving back to ext2 (no new hardware in the
budget for this area), I'm testing
this on another box.  Also looking at what might be tuneable in ext3
relative to memory usage.  
Any other advice is most welcome.  I'll post a summary once it fixed.

Thanks
Dave
>>> "Dave Warchol" <[email protected]> 12/24/03 07:27AM >>>
I forgot to mention that the Apache is built from source.

Hello:
     I have a box running Red Hat 9 (upgrade from 7.3), kernel
2.4.20-9smp, Apache 1.3.28, mod_perl 1.29, mod_ssl 2.8.15, OpenSSL
0.9.7a, PostgreSQL 7.2.3-3.  Ever since the upgrade, it appears that
something is using and not releasing memory.  The box is a Dell 4300. 
We recently added memory to bring it up to 1GB.  Right now I'm looking
at the box and its fairly quiet time (only 5-7 httpd processes
running),
however, roughly 930000k memory is being used.   Stopping apache
releases some.  When the box is first booted, only around 250000k is
used, stays like that for a while (will increase/decrease with # of
httpd processes), but then reaches a point where it stops behaving as
it
should.  Any pointers about where to look would be appreciated.  I've
also shutdown any extraneous processes.

Thanks
Dave


Dave Warchol
Technical Consultant
Hartford Hospital
Infomation Services
80 Seymour Street
Hartford, CT 06102  USA

Voice:  860.545.4222
Cell:  860.250.7253
Fax:  860.545.3321
[email protected] 
submissions: LinuxManagers-35TzE1X9F6582KRnZfj+bdi2O/[email protected] 
subscribe/unsubscribe:
http://www.linuxmanagers.org/mailman/listinfo/linuxmanagers
_______________________________________________
LinuxManagers mailing list - http://www.linuxmanagers.org
submissions: LinuxManagers-35TzE1X9F6582KRnZfj+bdi2O/[email protected]
subscribe/unsubscribe: http://www.linuxmanagers.org/mailman/listinfo/linuxmanagers