Re: infx cleaners

"Mike Clark" <mike_clark@wawb.uscourts.gov> wrote in message news:40A5118E.EB6D5D07@wawb.uscourts.gov... > A while back, we were told that a sign of an ill-configured infx > instance was a number of under-utilized cleaner threads. So we've been > tracking them and normally after a restart they all get busy. Now, > however, > > cleaners_5e738500: 23443 > cleaners_5e7389ec: 72185 > cleaners_5e738ed8: 136744 > cleaners_5e7393c4: 14071 > cleaners_5e7398b0: 5872 > cleaners_5e739d9c: 5620 > cleaners_5e73a288: 25511 > cleaners_5e73a774: 30844 > cleaners_5e73ac60: 25476 > cleaners_5e73b14c: 24600 > cleaners_5e73b638: 277 > cleaners_5e73bb24: 23386 > cleaners_5e73c010: 279 > cleaners_5e73c4fc: 14539 > cleaners_5e73c9e8: 23941 > cleaners_5e73ced4: 27691 > > There are 2 threads w/ just a couple hundred writes after 5 days uptime. > I forget the details - does anybody know what the story is w/ cleaner > threads and what I could/should be doing to keep them all busy? Make sure that you have LRUS=CLEANERS. Make sure chunks are created round-robin on the disks. Also since each cleaner is allocated one chunk to clean sometimes if the dirty pages are not spread across at least <CLEANERS> number of chunks then not all the cleaners get activated. It could just be that the spread of dirty pages is different at this time hence some of the cleaners are not active. Beside you should have some quiet cleaners some of the time, how else do you handle peaks loads of page cleaning? > >