On Thu, Jan 22, 2009 at 6:46 PM, Vadim Tkachenko <va...@percona.com> wrote: > Mark, > > we tested InnoDB with thread pool and it just hangs if count of > connection making transaction more than slots in pool. > > I think this is problem is impossible to resolve, as pool should know > about transaction state, which is very hard to do. >
We have not done any read-write tests with it. For fun, someone created 8000 connection on 100 threads. But this is something we need to think about. > > > MARK CALLAGHAN wrote: >> Do you have the thread pool code in your tree >> (--thread_handling=pool-of-threads)? >> Have you done anything to make it efficient? >> >> For me, sysbench OLTP readonly is up to 3X faster using blackhole when >> pool-of-threads is not used -- http://bugs.mysql.com/bug.php?id=42288, >> and this is on a 4 core server so I assume the overhead gets much >> worse with more cores. >> > > > -- > Vadim Tkachenko, CTO > Percona Inc. > ICQ: 369-510-335, Skype: vadimtk153, Phone +1-888-401-3403 > MySQL Performance Blog - http://www.mysqlperformanceblog.com > MySQL Consulting http://www.percona.com/ > -- Mark Callaghan mdcal...@gmail.com _______________________________________________ Mailing list: https://launchpad.net/~drizzle-discuss Post to : drizzle-discuss@lists.launchpad.net Unsubscribe : https://launchpad.net/~drizzle-discuss More help : https://help.launchpad.net/ListHelp
