After tuning Postgresql, PgBench results are worse -

-

i testing postgresql on 8gb ram/4 cpus/ 80gb ssd cloud server digital ocean. ran pgbench default settings in postgresql.conf, , altered common settings--shared_buffers, work_mem, maintenance_work_mem, effective_cache_size--to reflect 8gb of ram. after running 2nd set of tests, noticed of results worse. suggestions on why might be? rather new pgbench , tuning postgresql in general.

settings:

  • shared_buffers = 2048mb
  • work_mem = 68mb
  • maintenance_work_mem = 1024mb
  • effective_cache_size = 4096mb

tests:

  • pgbench -i -s 100
  • pgbench -c 16 -j 2 -t 60 -u postgres postgres
  • pgbench -s -c 16 -j 2 -t 60 -u postgres postgres
  • pgbench -c 16 -j 4 -t 60 -u postgres postgres
  • pgbench -s -c 16 -j 4 -t 60 -u postgres postgres
  • pgbench -c 16 -j 8 -t 60 -u postgres postgres
  • pgbench -s -c 16 -j 8 -t 60 -u postgres postgres

how effective these tests? effective way employ pgbench? how should customize tests reflect data , server instance?

what mean "worse"? how long time run pgbench? test should executed 2hour minimum realistic values. version of postgresql have?

attention: should careful interpretation pgbench result. should optimize execution of application, not pgbench. pgbench hw or sw checking, bad tool optimizing of postgresql configuration.

a mentioned configuration variables basic configuration , cannot wrong there (server must not use swap actively ever - , these variables ensure it).

a formula use:

 -- dedicated server 8gb ram shared_buffers = 1/3 .. 1/4 dedicated ram effecttive_cache_size = 2/3 dedicated ram  maintenance_work_mem > higher big table (if possible)                        else 1/10 ram                        else max_connection * 1/4 * work_mem  work_mem  = precious setting based on slow query analyse              (first setting 100mb)  --must true max_connection * work_mem * 2 + shared_buffers            + 1gb (o.s.) + 1gb (filesystem cache) <= ram size

usually default values of wal buffer size , checkpoint segments low too. , can increase it.


Comments

Post a Comment

Popular posts from this blog

java - Run a .jar on Heroku -

-
i trying run .jar file on heroku, , can't find instructions anywhere on how this. have read wants me compile uploading heroku. possible, , if is, how this? if helps, trying run minecraft craftbukkit server, , don't have pom.xml file yet. you didn't provide details i'll try give , people reaches question way it. first of all, create heroku account , create heroku application (assumed have done that). second, need generate application jar , upload heroku. so, open terminal , install heroku cli (assuming unix terminal): $ sudo npm install -g heroku-cli you need have npm , nodejs installed, can check link: https://www.digitalocean.com/community/tutorials/how-to-install-node-js-on-ubuntu-16-04 once done that, can use npm update node version acceptable heroku-cli: $ sudo npm install -g n $ sudo n 7.10.0 and run: $ sudo npm install -g heroku-cli in order deploy jars, need install deploy plugin: $ sudo heroku plugins:install heroku-cli-deploy ...
Read more

java - Jtable duplicate Rows -

-
Image
when add new rows in jtable duplicate rows many times this method update jtable public final void tableaucomande(){ ((defaulttablemodel)tabcommande.getmodel()).getdatavector().clear(); session s=hibernateutil.getsessionfactory().opensession(); criteria cr1= s.createcriteria(commande.class); list <commande> lcs=cr1.list() ; criteria cr2= s.createcriteria(lignecommande.class); list <lignecommande> lck=cr2.list() ; if(lcs.size()!=0 && lck.size()!=0) { (commande lc:lcs) { for(lignecommande lig:lck){ ((defaulttablemodel)tabcommande.getmodel()).addrow(new object[]{ lc.getid_cmd(),lc.getnumcmd(),lc.getdate_cmd(),lc.getclient().getnomprenom(), lig.getproduit().getlib_prd(),lig.getquantite(),lig.getproduit().getid_prd() }); } } // tabcommande.setrowselectioninterval(tabcommande.getrowcount()-1,tabcommande.getrowcount()-1); } } this code add add data rows displaye...
Read more

validation - How to pass paramaters like unix into windows batch file -

-
i need pass various parameters .cmd file unix format file.cmd -configuration=value -source=value -flag but, try this: startlocal @echo off cls setlocal set cmdline=%* set configuration= set source= set badargs= set validation= goto main :splitargs echo splitargs(%*) if "%*" neq "" ( /f "tokens=1,2,* delims== " %%i in ("%*") call :assignkeyvalue %%i %%j & call :splitargs %%k ) goto :eof :assignkeyvalue echo assignkeyvalue(%1, %2) if /i %1==-configuration ( set configuration=%2 ) else if /i %1==-source ( set source=%2 ) else ( rem append unrecognised [key,value] badargs echo unknown key %1 set badargs=%badargs%[%1, %2] ) goto :eof :validate echo validating set validation=fail if defined configuration ( echo -configuration ok if defined source ( echo -source ok if not defined badargs ( set validation=success ) ) ) goto :eof :main cl...
Read more