Re: [Maria-developers] Working on microsecond patches

2009-04-08 Thread MARK CALLAGHAN
On Wed, Apr 8, 2009 at 9:13 AM, Michael Widenius wrote: > > Hi! > >> "MARK" == MARK CALLAGHAN writes: > > MARK> On Wed, Mar 18, 2009 at 7:23 AM, Michael Widenius > wrote: >>> >>> Hi! >>> "Vadim" == Vadim Tkachenko writes: >>> > Vadim> Michael, > Vadim> Thread-pool should be used v

Re: [Maria-developers] Working on microsecond patches

2009-04-08 Thread Michael Widenius
Hi! > "MARK" == MARK CALLAGHAN writes: MARK> On Wed, Mar 18, 2009 at 7:23 AM, Michael Widenius wrote: >> >> Hi! >> >>> "Vadim" == Vadim Tkachenko writes: >> Vadim> Michael, Vadim> Thread-pool should be used very carefully. >> Vadim> We tested it and InnoDB hangs in sysbench bench

Re: [Maria-developers] Working on microsecond patches

2009-04-08 Thread Michael Widenius
Hi! > "Jeremiah" == Jeremiah Gowdy writes: Vadim> Michael, Vadim> Thread-pool should be used very carefully. >> Vadim> We tested it and InnoDB hangs in sysbench benchmarks when count of Vadim> client connection > size of thread-pool. >> >> Yes, this a problem when you have more locks than

Re: [Maria-developers] Working on microsecond patches

2009-03-18 Thread MARK CALLAGHAN
On Wed, Mar 18, 2009 at 7:23 AM, Michael Widenius wrote: > > Hi! > >> "Vadim" == Vadim Tkachenko writes: > > Vadim> Michael, > Vadim> Thread-pool should be used very carefully. > > Vadim> We tested it and InnoDB hangs in sysbench benchmarks when count of > Vadim> client connection > size of t

Re: [Maria-developers] Working on microsecond patches

2009-03-18 Thread Michael Widenius
Hi! > "Vadim" == Vadim Tkachenko writes: Vadim> Michael, Vadim> Thread-pool should be used very carefully. Vadim> We tested it and InnoDB hangs in sysbench benchmarks when count of Vadim> client connection > size of thread-pool. Yes, this a problem when you have more locks than threads. T

Re: [Maria-developers] Working on microsecond patches

2009-03-13 Thread Vadim Tkachenko
Michael, Thread-pool should be used very carefully. We tested it and InnoDB hangs in sysbench benchmarks when count of client connection > size of thread-pool. The problem is transaction state. Some connections may do internals locks and after that could not enter to pool, because all slots are

[Maria-developers] Working on microsecond patches

2009-03-13 Thread Michael Widenius
Hi! Just wanted to inform that I just pushed and extended pool-of-threads handling to the active MariaDB tree at launchpad. http://askmonty.org/wiki/index.php/MariaDB_versus_MySQL#Pool_of_threads_.28limited_sets_of_threads_handling_all_queries.29 I plan to during the next few days also add the