Mario Olimpio de Menezes wrote: > On Friday 24 February 2006 08:57, you wrote: >> Mario Olimpio de Menezes wrote: >>> On Thursday 23 February 2006 19:07, you wrote: [...] >> right. How does the RAM looks like on the server currently ? > > $ cat /proc/meminfo > MemTotal: 240696 kB > MemFree: 5004 kB > Buffers: 9956 kB > Cached: 41452 kB > SwapCached: 3704 kB > Active: 170056 kB > Inactive: 24676 kB > HighTotal: 0 kB > HighFree: 0 kB > LowTotal: 240696 kB > LowFree: 5004 kB > SwapTotal: 682720 kB > SwapFree: 639556 kB
Yes you don't have a lot a free mem still available. which is what I expected. > $ ps auxc | grep -E "zope|www-data" > zope 2450 0.0 13.4 109284 32264 ? S< Feb23 0:08 python2.3 > www-data 2578 0.0 5.4 37768 13160 ? S< Feb23 0:07 apache2 > www-data 2579 0.0 6.0 39048 14528 ? S< Feb23 0:14 apache2 > www-data 2580 0.0 5.3 37348 12856 ? S< Feb23 0:05 apache2 > www-data 2581 0.0 5.2 37112 12720 ? S< Feb23 0:05 apache2 > www-data 2582 0.0 6.8 41940 16560 ? S< Feb23 0:08 apache2 > www-data 2584 0.0 5.2 37056 12652 ? S< Feb23 0:08 apache2 > www-data 2599 0.0 7.4 42400 17832 ? S< Feb23 0:14 apache2 > www-data 2711 0.0 5.1 37056 12472 ? S< Feb23 0:03 apache2 > www-data 2712 0.0 5.3 37364 12896 ? S< Feb23 0:03 apache2 > www-data 2713 0.0 5.1 37040 12516 ? S< Feb23 0:03 apache2 > >> Let's start with this : >> >> - Add a cache on Apache. (or use SQUID if you prefer) > > using ProxyPass and ProxyPassReverse? (Apache) I'm talking about this : http://httpd.apache.org/docs/2.0/mod/mod_cache.html But you can do this in a second step. I'm not an Apache expert so I'll let others give you a hand on this :;) > > If I use Squid I shouldn't use Apache in this same machine, right? This might be possible. >> - Check the RAM and let's tune the ZODB caches after that. > > could you give me some more info on how to do this? yes. The default number of object per cache is 5000. This is too much for your configuration I think. Note, here each thread has its own cache so you need to multiply by 2 this number if you change your number of threads to 2. Try to decrease this number by changing the directive as below within your zope.conf : <zodb_db main> # Main FileStorage database cache-size 2000 <filestorage> path $INSTANCE/var/Data.fs </filestorage> mount-point / </zodb_db> You need to reboot your zope instance after this. You can check the status of your caches within the Zope control panel in ZMI : http://yourhost:PORT/Control_Panel/Database/main/manage_cacheParameters After that perform some tests and check how the number of objects within the cache evolved. If they are full and you still get some mem available then try to increase the cache size until you reach the limit. You'll be able to tune when your users will be using the portal. At a certain time you'll be limited by the mem of the server avilable of course. Note the bigger the cache size are the best the performances will be, of course. Tuning caches in Zope are a complex "cuisine" where you won't be able to find a single solution for every case. You need to try some configuration out and then adjust for you own server. >> you said you had 2 threads on the server right ? >> > no, I don't. where can I check this? etc/zope.conf within your Zope instance. Change the the zserver-thread directive. I'd change it to 2. > I will try to use ProxyPass and ProxyPassReverse to do Cache on Apache side oki. Keep us posted. J. -- Julien Anguenot | Nuxeo R&D (Paris, France) CPS Platform : http://www.cps-project.org Zope3 / ECM : http://www.z3lab.org mail: anguenot at nuxeo.com; tel: +33 (0) 6 72 57 57 66
signature.asc
Description: OpenPGP digital signature
_______________________________________________ cps-users mailing list [email protected] http://lists.nuxeo.com/mailman/listinfo/cps-users
