rel_3_0, zebra 2.0 & updating Koha through ZOOM (or : high load tests at SAN-OP)
Hi, The SAN-OP did some high load tests this week, with all librarians (around 100 computers) working with rel_3_0... Just to remember : 300 000 items, 7 branches. Here are some results : - with mod_perl activated & innoDB database for mySQL = good response times, some major problems on tuesday, solved on wednesday, everything OK on thursday - without mod_perl & innoDB database for mySQL = (really) too slow to be usable. - without mod_perl & myISAM database for mySQL = good response times, but we lack advanced features from innoDB (like foreign keys... + if I don't mind, myISAM is no more the default DB under mySQL 5.0) - they worked on zebra 2.0, with zebra real-time updates activated (zebraop NOT commented in Biblio.pm). The bad news : they face the same behaviour as joshua with zebra 1 :-((((( So, our friends from Indexdata still can't insure that we can use ZOOM to update biblios... - as a reminder : the 3.0 french translation is almost done, but pascale Nalon, that works on it has found some inconsistency in OPAC strings (like book bag & basket both used) The move to production will be on january, 3rd, keep fingers crossed... cheers -- Paul POULAIN et Henri Damien LAURENT Consultants indépendants en logiciels libres et bibliothéconomie (http://www.koha-fr.org) Tel : 04 91 31 45 19
On Fri, Dec 15, 2006 at 06:16:03PM +0100, Paul POULAIN wrote:
Just to remember : 300 000 items, 7 branches.
Here are some results : - with mod_perl activated & innoDB database for mySQL = good response times, some major problems on tuesday, solved on wednesday, everything OK on thursday
- without mod_perl & innoDB database for mySQL = (really) too slow to be usable. This is baffling to me. NPL has the exact same stats and hasn't been running mod_perl and Koha's faster than ever. Is there something we're missing that happened between dev_week and rel_3_0?
I would highly recommend taking a very hard look at where the bottleneck is without mod_perl. Here's where I'd start: $ perl -d:DProf -I /build/koha circulation.pl findborrower=0054259 tmon.out $ dprofpp -v > dprof.txt (using of course valid data from their dataset and other operations) Could you report back on the results of this test?
- without mod_perl & myISAM database for mySQL = good response times, but we lack advanced features from innoDB (like foreign keys... + if I don't mind, myISAM is no more the default DB under mySQL 5.0) This too is surprising. I suspect the problem is that the innodb configuration in my.cnf needs to be tweaked.
- they worked on zebra 2.0, with zebra real-time updates activated (zebraop NOT commented in Biblio.pm). The bad news : they face the same behaviour as joshua with zebra 1 :-((((( So, our friends from Indexdata still can't insure that we can use ZOOM to update biblios... Yea, this isn't a big surprise. I'd be very curious to know if you have the same problems with the latest 1.3. NPL's index hasn't crashed now for about a month and a half or so since the latest update.
- as a reminder : the 3.0 french translation is almost done, but pascale Nalon, that works on it has found some inconsistency in OPAC strings (like book bag & basket both used) Cool, pascale++
Cheers, -- Joshua Ferraro SUPPORT FOR OPEN-SOURCE SOFTWARE President, Technology migration, training, maintenance, support LibLime Featuring Koha Open-Source ILS jmf@liblime.com |Full Demos at http://liblime.com/koha |1(888)KohaILS
participants (2)
-
Joshua Ferraro -
Paul POULAIN