Greg, Good detective work. That is related, however we couldn't determine the exact connection. We do see a reduced occurrence of the waits with a lesser number of rollback segments, however when the rollback segments grew again to the thousands, the issue was not appearing nowhere as often as it did before. So we don't really have a conclusion, other than it's related. On Tue, Jun 8, 2010 at 1:48 AM, Greg Rahn <greg@xxxxxxxxxxxxxxxxxx> wrote: > Am I correct in assuming this is the findings of your issue? > > http://afatkulin.blogspot.com/2010/06/row-cache-objects-latch-contention.html > > On Wed, Jun 2, 2010 at 1:04 PM, Christo Kutrovsky > <kutrovsky.oracle@xxxxxxxxx> wrote: > > I am having a weird issue on Oracle 11.2 64 bit SPARC with T2+ CPUs. > > > > Occasionally my loading process (bulk insert statement) will have a large > > number of consecutive waits (as reported by ASH) of "latch: row cache > > objects". The statement is not been reparsed, and ASH reports > > "IN_SQL_EXECUTION". > > > > I do have "blocked by" information, and that is pointing to a parallel > merge > > statement been executed at the same time. The blockee also shows "N" for > all > > "IN_xx" columns in ASH. The > > > > Does anyone know how to get more information for which objects from the > > rowcache I am contending, and why at all? > > > > I am suspecting it's the parallel merge that is affecting the system > > globally, as a lot of queries just "don't run" during these short > freezes. > > -- > Regards, > Greg Rahn > http://structureddata.org > -- Christo Kutrovsky Senior Consultant Pythian.com I blog at http://www.pythian.com/blogs/