long checkpoint problem

Hello! Sorry my English. There is problem with checkpoint duration. IDS 7.31. Facts: - user run mass inserts in temp table (with no log) - there are 3 indexes on this temp table - i see very large numbers of dirty buffers. IDS makes lru-writes but buffers are dirties faster. Result is long checkpoint. - in use LRU_MAX_DIRTY 2 LRU_MIN_DIRTY 1 LRUS 128 PHYSFILE 20300 NUMCPUVPS 2 CLEANERS 2 BUFFERS 440000 CKPTINTVL 300 Does really indexes builds in resident portion SHM? What can I do to reduce checkpoint duration? Is it safe to reduce physical log size? What if overflow occured? -- Best regards, Alex

It's quite plausible that page cleaning would lag behind buffer writes during a bulk load. Disk i/o is obviously inherently much, much slower than buffer i/o. It's likely to be slower if: - Everything's being written out to one disk or volume (e.g. the table doesn't use fragmentation) - CLEANERS != no. of disks it's writing to - You're using RAID5 (NO RAID5! NO RAID5! NO RAID5!) - You're using INSERTs rather than HPL (though to use HPL you'd have to redefine the table as a permanent one) - AIO VPs aren't tuned correctly - The table hasn't been sized correctly so it keeps having to write out new extent allocation information Care to post some info on your disk config? What were you hoping to achieve by making the physical log smaller? If it was to increase checkpoint frequency then yes this would achieve that effect but could be lethal if you made it TOO small. Try adjusting CKPTINTVL instead. Andy spirin_alex@mail.ru (Alex V. Spirin) wrote in message news:<1c930d8c.0401282232.7e66d6b9@posting.google.com>... > Hello! > > Sorry my English. > > There is problem with checkpoint duration. IDS 7.31. > Facts: > - user run mass inserts in temp table (with no log) > - there are 3 indexes on this temp table > - i see very large numbers of dirty buffers. IDS makes lru-writes > but buffers are dirties faster. Result is long checkpoint. > - in use LRU_MAX_DIRTY 2 > LRU_MIN_DIRTY 1 > LRUS 128 > PHYSFILE 20300 > NUMCPUVPS 2 > CLEANERS 2 > BUFFERS 440000 > CKPTINTVL 300> > Does really indexes builds in resident portion SHM? > What can I do to reduce checkpoint duration? Is it safe to reduce > physical log size? What if overflow occured?