Hi Paul, thx a lot. So Im going to do the tests with 1.1 again. Or would you recommend 1.5?
Btw. PBXT already rocks (on my tests) vs xtradb doing massive non-PK-SELECTs. Thx for the great Work\o/ Erkan On Mon, Nov 29, 2010 at 10:23:38AM +0100, Paul McCullagh wrote: > Hi Erkan, > > On Nov 29, 2010, at 4:30 AM, erkan yanar wrote: > > > > >Playing around from time to time PBXT stops doing anything: > > > >xtstat: > > > >time xact row rec rec rec rec rec rec rec ind > >ind ind ind ind ind ilog ilog ilog xlog xlog > >xlog xlog xlog xlog data data data data to to to xact > >msec commt ins in out syncs/ms hits miss frees %use in > >out syncs/ms hits miss %use in out syncs/ms in out > >syncs msec hits miss in out syncs msec chkpt write sweep > >dirty > >5001 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/5001 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 1049M 400M 85.6M > >3865 > >5001 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/5001 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 1049M 400M 85.6M > >3865 > >5001 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/5001 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 1049M 400M 85.6M > >3865 > >5001 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/5001 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 1049M 400M 85.6M > >3865 > >5001 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/5001 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 1049M 400M 85.6M > >3865 > >5001 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/5001 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 1049M 400M 85.6M > >3865 > >5000 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/5000 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 1049M 400M 85.6M > >3865 > >5001 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/5001 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 1049M 400M 85.6M > >3865 > >5001 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/5001 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 1049M 400M 85.6M > >3865 > >5000 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/5000 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 1049M 400M 85.6M > >3865 > >5001 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/5001 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 1049M 400M 85.6M > >3865 > >5001 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/5001 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 1049M 400M 85.6M > >3865 > >5000 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/5000 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 1049M 400M 85.6M > >3865 > >5001 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/5001 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 1049M 400M 85.6M > >3865 > >5001 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/5001 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 1049M 400M 85.6M > >3865 > >5000 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/5000 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 1049M 400M 85.6M > >3865 > >5000 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/5000 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 1049M 400M 85.6M > >3865 > >5000 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/5000 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 1049M 400M 85.6M > >3865 > >5000 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/5000 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 1049M 400M 85.6M > >3865 > >5000 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/5000 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 1049M 400M 85.6M > >3865 > >5000 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/5000 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 1049M 400M 85.6M > >3865 > >5001 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/5001 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 1049M 400M 85.6M > >3865 > >5001 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/5001 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 1049M 400M 85.6M > >3865 > >5001 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/5001 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 1049M 400M 85.6M > >3865 > >5000 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/5000 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 1049M 400M 85.6M > >3865 > >time xact row rec rec rec rec rec rec rec ind > >ind ind ind ind ind ilog ilog ilog xlog xlog > >xlog xlog xlog xlog data data data data to to to xact > >msec commt ins in out syncs/ms hits miss frees %use in > >out syncs/ms hits miss %use in out syncs/ms in out > >syncs msec hits miss in out syncs msec chkpt write sweep > >dirty > >5000 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/5000 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 1049M 400M 85.6M > >3865 > >5001 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/5001 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 1049M 400M 85.6M > >3865 > >5000 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/5000 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 1049M 400M 85.6M > >3865 > >5001 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/5001 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 1049M 400M 85.6M > >3865 > >5000 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/5000 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 1049M 400M 85.6M > >3865 > >5000 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/5000 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 1049M 400M 85.6M > >3865 > >5000 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/5000 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 1049M 400M 85.6M > >3865 > >5000 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/5000 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 1049M 400M 85.6M > >3865 > >5000 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/5000 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 1049M 400M 85.6M > >3865 > >5001 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/5001 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 1049M 400M 85.6M > >3865 > >5000 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/5000 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 1049M 400M 85.6M > >3865 > >5000 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/5000 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 1049M 400M 85.6M > >3865 > >5000 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/5000 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 1049M 400M 85.6M > >3865 > >5000 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/5000 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 1049M 400M 85.6M > >3865 > >5000 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/5000 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 1049M 400M 85.6M > >3865 > >5000 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/5000 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 1049M 400M 85.6M > >3865 > >5000 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/5000 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 1049M 400M 85.6M > >3865 > >5000 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/5000 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 1049M 400M 85.6M > >3865 > >5001 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/5001 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 1049M 400M 85.6M > >3865 > > > > The problem here is the index cache is full, and the thread which is > flushing the index is hanging in sync (i.e. flush to disk). The rest > of the system is hanging because it needs a clean page in the index. > > I did quite a bit of work to improve this in 1.1 and 1.5. > > But in general, if the index cache is really large, it need to be > flushed more regularly to ensure that too many changes do not > accumulate. Checkpointing should do this, but there is a bug in 1.0. > > You checkpoint frequency is 128M, xtstat says to-chkpt = 1049M. This > means the checkpoint has not occurred as often as he should. > > This problem should be fixed in 1.1. > > >While I suggest there should be something done. > > > >Processlist: > > > >| 311367 | root | localhost | test | Query | 701 | > >Updating | UPDATE rein SET id3=round(rand()*10000) where i...@id | > >| 311366 | root | localhost | test | Query | 690 | > >Updating | UPDATE rein SET id3=round(rand()*10000) where i...@id | > >| 311368 | root | localhost | test | Query | 701 | > >Updating | UPDATE rein SET id3=round(rand()*10000) where i...@id | > >| 311369 | root | localhost | test | Query | 701 | > >Updating | UPDATE rein SET id3=round(rand()*10000) where i...@id | > >| 311372 | root | localhost | test | Query | 701 | > >Updating | UPDATE rein SET id3=round(rand()*10000) where i...@id | > >| 311373 | root | localhost | test | Query | 694 | > >Updating | UPDATE rein SET id3=round(rand()*10000) where i...@id | > >| 311376 | root | localhost | test | Query | 691 | > >Updating | UPDATE rein SET id3=round(rand()*10000) where i...@id | > >| 311377 | root | localhost | test | Query | 698 | > >Updating | UPDATE rein SET id3=round(rand()*10000) where i...@id | > >| 311379 | root | localhost | test | Query | 701 | > >Updating | UPDATE rein SET id3=round(rand()*10000) where i...@id | > >| 311380 | root | localhost | test | Query | 701 | > >Updating | UPDATE rein SET id3=round(rand()*10000) where i...@id | > >| 311382 | root | localhost | test | Query | 696 | > >Updating | UPDATE rein SET id3=round(rand()*10000) where i...@id | > >| 311384 | root | localhost | test | Query | 699 | > >Updating | UPDATE rein SET id3=round(rand()*10000) where i...@id | > >| 311385 | root | localhost | test | Query | 693 | > >Updating | UPDATE rein SET id3=round(rand()*10000) where i...@id | > >| 311388 | root | localhost | test | Query | 691 | > >Updating | UPDATE rein SET id3=round(rand()*10000) where i...@id | > >| 311389 | root | localhost | test | Query | 691 | > >Updating | UPDATE rein SET id3=round(rand()*10000) where i...@id | > >| 311390 | root | localhost | test | Query | 696 | > >Updating | UPDATE rein SET id3=round(rand()*10000) where i...@id | > >| 311391 | root | localhost | test | Query | 701 | > >Updating | UPDATE rein SET id3=round(rand()*10000) where i...@id | > >| 311392 | root | localhost | test | Query | 699 | > >Updating | UPDATE rein SET id3=round(rand()*10000) where i...@id | > >| 311395 | root | localhost | test | Query | 687 | > >Updating | UPDATE rein SET id3=round(rand()*10000) where i...@id | > >| 311399 | root | localhost | test | Query | 690 | > >Updating | UPDATE rein SET id3=round(rand()*10000) where i...@id | > >| 311400 | root | localhost | test | Query | 701 | > >Updating | UPDATE rein SET id3=round(rand()*10000) where i...@id | > >| 311404 | root | localhost | test | Query | 694 | > >Updating | UPDATE rein SET id3=round(rand()*10000) where i...@id | > >| 311405 | root | localhost | test | Query | 690 | > >Updating | UPDATE rein SET id3=round(rand()*10000) where i...@id | > >| 311406 | root | localhost | test | Query | 694 | > >Updating | UPDATE rein SET id3=round(rand()*10000) where i...@id | > >| 311407 | root | localhost | test | Query | 699 | > >Updating | UPDATE rein SET id3=round(rand()*10000) where i...@id | > >| 311408 | root | localhost | test | Query | 691 | > >Updating | UPDATE rein SET id3=round(rand()*10000) where i...@id | > >| 311409 | root | localhost | test | Query | 688 | > >Updating | UPDATE rein SET id3=round(rand()*10000) where i...@id | > >| 311411 | root | localhost | test | Query | 694 | > >Updating | UPDATE rein SET id3=round(rand()*10000) where i...@id | > >| 311412 | root | localhost | test | Query | 694 | > >Updating | UPDATE rein SET id3=round(rand()*10000) where i...@id | > >| 311414 | root | localhost | test | Query | 701 | > >Updating | UPDATE rein SET id3=round(rand()*10000) where i...@id | > >| 311416 | root | localhost | test | Query | 696 | > >Updating | UPDATE rein SET id3=round(rand()*10000) where i...@id | > >| 311419 | root | localhost | test | Query | 691 | > >Updating | UPDATE rein SET id3=round(rand()*10000) where i...@id | > >| 311422 | root | localhost | test | Query | 687 | > >Updating | UPDATE rein SET id3=round(rand()*10000) where i...@id | > >| 311423 | root | localhost | test | Query | 690 | > >Updating | UPDATE rein SET id3=round(rand()*10000) where i...@id | > >| 311424 | root | localhost | test | Query | 696 | > >Updating | UPDATE rein SET id3=round(rand()*10000) where i...@id | > >| 311425 | root | localhost | test | Query | 691 | > >Updating | UPDATE rein SET id3=round(rand()*10000) where i...@id | > >| 311426 | root | localhost | test | Query | 691 | > >Updating | UPDATE rein SET id3=round(rand()*10000) where i...@id | > >| 311427 | root | localhost | test | Query | 689 | > >Updating | UPDATE rein SET id3=round(rand()*10000) where i...@id | > > > > > > > >After a while it starts doing some tings again: > >time xact row rec rec rec rec rec rec rec ind > >ind ind ind ind ind ilog ilog ilog xlog xlog > >xlog xlog xlog xlog data data data data to to to xact > >msec commt ins in out syncs/ms hits miss frees %use in > >out syncs/ms hits miss %use in out syncs/ms in out > >syncs msec hits miss in out syncs msec chkpt write sweep > >dirty > >5000 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/5000 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 1049M 400M 85.6M > >3865 > >5001 0 0 0 0 0/0 0 0 0 76.0 > >0 0 1/1307 0 0 100 0 2753M 2/1 0 > >0 0 0 0 0 0 0 0 0 888M 400M 85.6M > >3865 > >5001 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/0 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 888M 400M 85.6M > >3865 > >5001 0 0 0 40 3/1300 0 0 0 76.0 > >0 0 0/0 0 0 100 0 0 1/8714 0 > >0 0 0 0 0 0 0 0 0 888M 400M 85.6M > >3865 > >5001 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/0 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 888M 400M 85.6M > >3865 > >5001 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/0 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 888M 400M 85.6M > >3865 > >5001 0 0 0 0 0/0 0 0 0 76.0 0 > >2906M 0/474 0 0 100 2758M 9 2/24.4t 0 0 > >0 0 0 0 0 0 0 0 888M 400M 85.6M 3865 > > > >ca. 110x > >5001 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/5001 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 888M 400M 85.6M > >3865 > >then > > > >time xact row rec rec rec rec rec rec rec ind > >ind ind ind ind ind ilog ilog ilog xlog xlog > >xlog xlog xlog xlog data data data data to to to xact > >msec commt ins in out syncs/ms hits miss frees %use in > >out syncs/ms hits miss %use in out syncs/ms in out > >syncs msec hits miss in out syncs msec chkpt write sweep > >dirty > >5001 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/5001 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 888M 400M 85.6M > >3865 > >5001 8 0 0 38.7M 0/0 13.0t 2 0 76.0 > >98304 0 1/438 17.9t 6 100 0 274K 3/1 0 > >130K 2 3320 2587t 0 0 0 0 0 358M 296M > >85.7M 3871 > >5001 0 0 0 47.2M 0/0 0 0 0 76.0 > >0 0 0/0 0 0 100 0 0 0/0 0 > >0 0 5001 3027t 0 0 0 0 0 358M 166M 85.7M > >3871 > >5001 4 0 0 44.0M 0/0 6482 0 0 76.0 65536 > >288K 1/1 8888 4 100 274K 16.7M 4/10.2t 0 66560 > >3 852 2013t 0 0 0 0 0 358M 32.7M 85.8M 3874 > >5001 0 0 0 14.2M 0/0 0 0 0 76.0 > >0 0 0/0 0 0 100 0 0 0/0 0 > >0 0 0 852t 0 0 0 0 0 358M 5550 85.8M > >3874 > >5001 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/0 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 358M 5550 85.8M > >3874 > >5001 42 0 0 0 0/0 63.8t 9 0 76.0 400K > >89.3M 14/261 87.3t 25 100 84.8M 83.1M 56/17.4t 0 620K > >16 15 16 0 0 0 0 0 359M 617K 86.4M 3902 > >5001 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/0 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 359M 617K 86.4M > >3902 > >5001 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/0 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 359M 617K 86.4M > >3902 > >5001 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/0 0 0 100 0 0 0/0 0 > >0 0 0 0 0 0 0 0 0 359M 617K 86.4M > >3902 > >5001 30 0 0 0 0/0 1017t 5 0 76.0 176K > >70.8M 17/1489 2987t 11 100 67.3M 292M 68/17.3t 0 10.2M > >19 2321 20 0 0 0 0 0 361M 10.8M 69.3M 3104 > >5001 0 0 0 0 0/0 247t 0 0 76.0 > >0 0 0/0 741t 0 100 0 0 0/0 0 > >0 0 5001 0 0 0 0 0 0 361M 10.8M 64.9M > >2898 > >5001 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/0 0 0 100 0 0 0/0 0 > >0 0 5001 0 0 0 0 0 0 361M 10.8M 64.9M > >2898 > >5001 0 0 0 0 0/0 0 0 0 76.0 > >0 0 0/0 0 0 100 0 0 0/0 0 > >0 0 5001 0 0 0 0 0 0 361M 10.8M 64.9M > >2898 > >5001 0 0 0 0 0/0 1850t 0 0 76.0 > >65536 0 0/0 5557t 4 100 0 0 0/0 0 > >26.9M 21 4268 45.1t 0 0 0 0 0 382M 37.7M > >53.1M 1352 > >5001 0 0 0 0 0/0 964t 0 0 76.0 16384 > >254M 0/3361 2896t 1 100 241M 9 2/22.2t 0 14.0M > >5 3770 36.3t 0 0 0 0 0 394M 51.7M 47.4M 549 > >5001 0 0 0 0 0/0 377t 0 0 76.0 > >288K 0 0/5001 1145t 18 100 0 0 0/0 0 > >8192K 1 4993 1 0 0 0 0 0 400M 59.7M > >45.7M 234 > > This activity above is checkpointing. This pauses in-between where > there is not flushing and no other activity are a bug in 1.0, as far > as I can remember. > > >and even the processlist shinks: > >| 311395 | root | localhost | test | Query | 1781 | > >Updating | UPDATE rein SET id3=round(rand()*10000) where i...@id | > >| 311409 | root | localhost | test | Query | 1782 | > >Updating | UPDATE rein SET id3=round(rand()*10000) where i...@id | > >| 311419 | root | localhost | test | Query | 1785 | > >Updating | UPDATE rein SET id3=round(rand()*10000) where i...@id | > >| 311422 | root | localhost | test | Query | 1781 | > >Updating | UPDATE rein SET id3=round(rand()*10000) where i...@id | > >| 311429 | root | localhost | test | Sleep | 750 > >| | NULL | > > > > > >MariaDB [test]> show global variables like '%pbxt%'; > >+------------------------------+-------+ > >| Variable_name | Value | > >+------------------------------+-------+ > >| pbxt_auto_increment_mode | 0 | > >| pbxt_checkpoint_frequency | 128M | > >| pbxt_data_file_grow_size | 50M | > >| pbxt_data_log_threshold | 256M | > >| pbxt_flush_log_at_trx_commit | 0 | > >| pbxt_garbage_threshold | 50 | > >| pbxt_index_cache_size | 8G | > >| pbxt_log_buffer_size | 512M | > >| pbxt_log_cache_size | 1G | > >| pbxt_log_file_count | 3 | > >| pbxt_log_file_threshold | 32MB | > >| pbxt_max_threads | 1507 | > >| pbxt_offline_log_function | 0 | > >| pbxt_record_cache_size | 12G | > >| pbxt_row_file_grow_size | 10M | > >| pbxt_support_xa | ON | > >| pbxt_sweeper_priority | 2 | > >| pbxt_transaction_buffer_size | 8M | > >--->> Wildest guessing as I didn't found best practices on tuning > >them. > > The settings are fine. In 1.1 I would recommend decreasing (which > means using a higher value) the checkpoint frequency. But in 1.0 this > value is not correctly used anyway. > > >MariaDB [test]> select version(); > >+-------------------+ > >| version() | > >+-------------------+ > >| 5.2.3-MariaDB-log | > > > >MariaDB [test]> select * from information_schema.PLUGINS where > >PLUGIN_NAME='PBXT'\G > >*************************** 1. row *************************** > > PLUGIN_NAME: PBXT > > PLUGIN_VERSION: 0.1 > > PLUGIN_STATUS: ACTIVE > > PLUGIN_TYPE: STORAGE ENGINE > > PLUGIN_TYPE_VERSION: 50203.0 > > PLUGIN_LIBRARY: NULL > >PLUGIN_LIBRARY_VERSION: NULL > > PLUGIN_AUTHOR: Paul McCullagh, PrimeBase Technologies GmbH > > PLUGIN_DESCRIPTION: High performance, multi-versioning > >transactional engine > > PLUGIN_LICENSE: GPL > > PLUGIN_MATURITY: Gamma > > PLUGIN_AUTH_VERSION: 1.0.11-7 Pre-GA > > > > > ># free -m > > total used free shared buffers > >cached > >Mem: 96682 74052 22630 0 180 > >32567 > >-/+ buffers/cache: 41304 55378 > >Swap: 2047 0 2047 > > > >CPU: > >2xX5550 > > > >Im clueless, how to interpret this behavior > > Basically the system is filling up the cache and then hitting a wall. > All background threads are way behind, mostly because of this (I would > guess, assuming you have no long running transactions or SELECTs). > > Unfortunately, all the work on flushing behavior was done in 1.1 and > 1.5. For example, 1.1 include multiply flush threads for better > throughput. > > Best regards, > > Paul > > > > > -- über den grenzen muß die freiheit wohl wolkenlos sein _______________________________________________ Mailing list: https://launchpad.net/~pbxt-discuss Post to : pbxt-discuss@lists.launchpad.net Unsubscribe : https://launchpad.net/~pbxt-discuss More help : https://help.launchpad.net/ListHelp