We do not have such high update frequency. So we never encountered this problem. If it is possible to take the slave offline during auto warming that is a good solution. --Noble
On Thu, Apr 9, 2009 at 2:02 PM, sunnyfr <johanna...@gmail.com> wrote: > > Hi Noble, > > Yes exactly that, > I would like to know how people do during a replication ? > Do they turn off servers and put a high autowarmCount which turn off the > slave for a while like for my case, 10mn to bring back the new index and > then autowarmCount maybe 10 minutes more. > > Otherwise I tried to put large number of mergefactor but I guess I've too > much update every 30mn something like 2000docs and almost all segment are > modified. > > What would you reckon? :( .... :) > > Thanks a lot Noble > > > Noble Paul നോബിള് नोब्ळ् wrote: >> >> So what I decipher from the numbers is w/o queries Solr replication is >> not performing too badly. The queries are inherently slow and you wish >> to optimize the query performance itself. >> am I correct? >> >> On Tue, Apr 7, 2009 at 7:50 PM, sunnyfr <johanna...@gmail.com> wrote: >>> >>> Hi, >>> >>> So I did two test on two servers; >>> >>> First server : with just replication every 20mn like you can notice: >>> http://www.nabble.com/file/p22930179/cpu_without_request.png >>> cpu_without_request.png >>> http://www.nabble.com/file/p22930179/cpu2_without_request.jpg >>> cpu2_without_request.jpg >>> >>> Second server : with one first replication and a second one during query >>> test: between 15:32pm and 15h41 >>> during replication (checked on .../admin/replication/index.jsp) my >>> respond >>> time query at the end was around 5000msec >>> after the replication I guess during commitment I couldn't get answer of >>> my >>> query for a long time, I refreshed my page few minutes after. >>> http://www.nabble.com/file/p22930179/cpu_with_request.png >>> cpu_with_request.png >>> http://www.nabble.com/file/p22930179/cpu2_with_request.jpg >>> cpu2_with_request.jpg >>> >>> Now without replication I kept going query on the second server, and I >>> can't >>> get better than >>> 1000msec repond time and 11request/second. >>> http://www.nabble.com/file/p22930179/cpu_.jpg cpu_.jpg >>> >>> This is my request : >>> select?fl=id&fq=status_published:1+AND+status_moderated:0+AND+status_personal:0+AND+status_private:0+AND+status_deleted:0+AND+status_error:0+AND+status_ready_web:1&json.nl=map&wt=json&start=0&version=1.2&bq=status_official:1^1.5+OR+status_creative:1^1+OR+language:en^0.5&bf=recip(rord(created),1,10,10)^3+pow(stat_views,0.1)^15+pow(stat_comments,0.1)^15&rows=100&qt=dismax&qf=title_en^0.8+title^0.2+description_en^0.3+description^0.2+tags^1+owner_login^0.5 >>> >>> Do you have advice ? >>> >>> Thanks Noble >>> >>> >>> -- >>> View this message in context: >>> http://www.nabble.com/solr-1.4-memory-jvm-tp22913742p22930179.html >>> Sent from the Solr - User mailing list archive at Nabble.com. >>> >>> >> >> >> >> -- >> --Noble Paul >> >> > > -- > View this message in context: > http://www.nabble.com/solr-1.4-memory-jvm-tp22913742p22966630.html > Sent from the Solr - User mailing list archive at Nabble.com. > > -- --Noble Paul
