RE: Long checkpoints

Ideas embedded >===== Original Message From alanolya@planmatics.com ===== >Hi, > >I would welcome suggestions on how to lower the checkpoint durations at >one of our sites. Relevant ONCONFIG parameters are given below. All >chunks are cooked files. At checkpoint time, there are on the average >about 500 dirty buffers to flush. From what I have heard going to raw space might cut checkpoints down by 20% > >Thanks. > >Alanoly Andrews. > >====================================== >Informix Dynamic Server 7.31 UC3 on AIX 4.3.2 > >----------------------------------- >Here are the last 20 lines of the Server Log: > >10:31:34 Checkpoint Completed: duration was 10 seconds. >10:36:46 Checkpoint Completed: duration was 11 seconds. >10:41:55 Checkpoint Completed: duration was 10 seconds. >10:47:07 Checkpoint Completed: duration was 11 seconds. >10:52:15 Checkpoint Completed: duration was 9 seconds. >10:57:25 Checkpoint Completed: duration was 9 seconds. >11:02:35 Checkpoint Completed: duration was 9 seconds. >11:05:16 Logical Log 161 Complete. >11:05:16 Logical Log 161 - Backup Started >11:05:19 Logical Log 161 - Backup Completed >11:07:46 Checkpoint Completed: duration was 11 seconds. >11:12:55 Checkpoint Completed: duration was 9 seconds. >11:18:04 Checkpoint Completed: duration was 9 seconds. >11:23:13 Checkpoint Completed: duration was 8 seconds. >11:28:25 Checkpoint Completed: duration was 11 seconds. >11:33:33 Checkpoint Completed: duration was 9 seconds. >11:38:53 Checkpoint Completed: duration was 19 seconds. >11:44:03 Checkpoint Completed: duration was 9 seconds. >11:49:10 Checkpoint Completed: duration was 8 seconds. >11:54:21 Checkpoint Completed: duration was 10 seconds.> >-------------------------------------- > >Relevant parts of the "onconfig" file: > >MULTIPROCESSOR 0 # 0 for single-processor, 1 for multi->processor >NUMCPUVPS 1 # Number of user (cpu) vps >SINGLE_CPU_VP 1 # If non-zero, limit number of cpu vps>to one > >NOAGE 1 # Process aging >AFF_SPROC 0 # Affinity start processor >AFF_NPROCS 0 # Affinity number of processors> ># Shared Memory Parameters > >LOCKS 150000 # Maximum number of locks >BUFFERS 45000 # Maximum number of shared buffers >NUMAIOVPS 2 # Number of IO vps I imagine this is to low. The advice usually given on this is as follows do onstat -g iov look at the io/wup column. If one of the aio has a value below 1 I imagine you have enough. Do not do this test till after increasing LRU's and CLEANERS. >PHYSBUFF 64 # Physical log buffer size (Kbytes) >LOGBUFF 64 # Logical log buffer size (Kbytes)>LOGSMAX 100 # Maximum number of logical log files >CLEANERS 2 # Number of buffer cleaner processes I recommend this be the same as LRU's. Cleaners flush your LRU queues. >SHMBASE 0x30000000 # Shared memory base address >SHMVIRTSIZE 100000 # initial virtual shared memory segment>size >SHMADD 64000 # Size of new shared memory segments >(Kbytes) >SHMTOTAL 720000 # Total shared memory (Kbytes). >0=>unlimited >CKPTINTVL 300 # Check point interval (in sec) >LRUS 6 # Number of LRU queues I would put this up at 50. This splits your buffers up into shorter LRU queus.. >LRU_MAX_DIRTY 10 # LRU percent dirty begin cleaning limit >LRU_MIN_DIRTY 5 # LRU percent dirty end cleaning limit If you only have 500 dirty buffers changing these wont help >LTXHWM 50 # Long transaction high water mark>percentage >LTXEHWM 60 # Long transaction high water mark >(exclusive) >TXTIMEOUT 0x12c # Transaction timeout (in sec) >STACKSIZE 64 # Stack size (Kbytes)> Hope these suggestions help. Will ------------------------------------------------------------ This e-mail has been sent to you courtesy of OperaMail, as a free service from Opera Software, makers of the award-winning Web Browser, Opera. Visit us at http://www.opera.com/ or our portal at: http://www.myopera.com/ Your free e-mail account is waiting at: http://www.operamail.com/ ------------------------------------------------------------

In article <8fhopu$c5n$1@news.xmission.com>, William Rice <ricew@operamail.com> writes > >Ideas embedded >>===== Original Message From alanolya@planmatics.com ===== >>Hi, >> >>I would welcome suggestions on how to lower the checkpoint durations at >>one of our sites. Relevant ONCONFIG parameters are given below. All >>chunks are cooked files. At checkpoint time, there are on the average >>about 500 dirty buffers to flush. > >From what I have heard going to raw space might cut checkpoints down by 20% > Arrgh! Do not decrease LRU_MIN_DIRTY and LRU_MAX_DIRTY to 1 and 0! This will probably decrease %wcache in onstat -p output to < 85%. Instead may increase LRUS to 64 and CLEANERS to 32 We run with LRUS 127 and CLEANERS 48 and have reduced checkpoint times from up to 20-120 seconds down to a 1-8 seconds. >> >>Thanks. >> >>Alanoly Andrews. >> >>====================================== >>Informix Dynamic Server 7.31 UC3 on AIX 4.3.2 >> >>----------------------------------- >>Here are the last 20 lines of the Server Log: >> >>10:31:34 Checkpoint Completed: duration was 10 seconds. >>10:36:46 Checkpoint Completed: duration was 11 seconds. >>10:41:55 Checkpoint Completed: duration was 10 seconds. >>10:47:07 Checkpoint Completed: duration was 11 seconds. >>10:52:15 Checkpoint Completed: duration was 9 seconds. >>10:57:25 Checkpoint Completed: duration was 9 seconds. >>11:02:35 Checkpoint Completed: duration was 9 seconds. >>11:05:16 Logical Log 161 Complete. >>11:05:16 Logical Log 161 - Backup Started >>11:05:19 Logical Log 161 - Backup Completed >>11:07:46 Checkpoint Completed: duration was 11 seconds. >>11:12:55 Checkpoint Completed: duration was 9 seconds. >>11:18:04 Checkpoint Completed: duration was 9 seconds. >>11:23:13 Checkpoint Completed: duration was 8 seconds. >>11:28:25 Checkpoint Completed: duration was 11 seconds. >>11:33:33 Checkpoint Completed: duration was 9 seconds. >>11:38:53 Checkpoint Completed: duration was 19 seconds. >>11:44:03 Checkpoint Completed: duration was 9 seconds. >>11:49:10 Checkpoint Completed: duration was 8 seconds. >>11:54:21 Checkpoint Completed: duration was 10 seconds.>> >>-------------------------------------- >> >>Relevant parts of the "onconfig" file: >> >>MULTIPROCESSOR 0 # 0 for single-processor, 1 for multi->>processor >>NUMCPUVPS 1 # Number of user (cpu) vps >>SINGLE_CPU_VP 1 # If non-zero, limit number of cpu vps>>to one >> >>NOAGE 1 # Process aging >>AFF_SPROC 0 # Affinity start processor >>AFF_NPROCS 0 # Affinity number of processors>> >># Shared Memory Parameters >> >>LOCKS 150000 # Maximum number of locks >>BUFFERS 45000 # Maximum number of shared buffers >>NUMAIOVPS 2 # Number of IO vps> >I imagine this is to low. The advice usually given on this is as follows >do onstat -g iov >look at the io/wup column. If one of the aio >has a value below 1 I imagine you have enough. >Do not do this test till after increasing LRU's and CLEANERS. > >>PHYSBUFF 64 # Physical log buffer size (Kbytes) >>LOGBUFF 64 # Logical log buffer size (Kbytes)>>LOGSMAX 100 # Maximum number of logical log files >>CLEANERS 2 # Number of buffer cleaner processes> >I recommend this be the same as LRU's. Cleaners flush your LRU queues. > >>SHMBASE 0x30000000 # Shared memory base address >>SHMVIRTSIZE 100000 # initial virtual shared memory segment>>size >>SHMADD 64000 # Size of new shared memory segments >>(Kbytes) >>SHMTOTAL 720000 # Total shared memory (Kbytes). >>0=>unlimited >>CKPTINTVL 300 # Check point interval (in sec) >>LRUS 6 # Number of LRU queues> >I would put this up at 50. This splits your buffers up into shorter LRU >queus.. > >>LRU_MAX_DIRTY 10 # LRU percent dirty begin cleaning limit >>LRU_MIN_DIRTY 5 # LRU percent dirty end cleaning limit> >If you only have 500 dirty buffers changing these wont help > >>LTXHWM 50 # Long transaction high water mark>>percentage >>LTXEHWM 60 # Long transaction high water mark >>(exclusive) >>TXTIMEOUT 0x12c # Transaction timeout (in sec) >>STACKSIZE 64 # Stack size (Kbytes)>> > >Hope these suggestions help. > >Will > >------------------------------------------------------------ >This e-mail has been sent to you courtesy of OperaMail, as a free service >from >Opera Software, makers of the award-winning Web Browser, Opera. Visit us at >http://www.opera.com/ or our portal at: http://www.myopera.com/ Your free e-mail >account is waiting at: http://www.operamail.com/ >------------------------------------------------------------ > -- David Williams

I would be careful with some of these parameters on a single CPU machine. My advice would be to increase the checkpoint duration to 600, and lower the MAX and MIN to 3 and 2. Adding cleaners and LRU's may help, but might not have the impact on a single CPU system that you desire. David Williams wrote: > In article <8fhopu$c5n$1@news.xmission.com>, William Rice > <ricew@operamail.com> writes > > > >Ideas embedded > >>===== Original Message From alanolya@planmatics.com ===== > >>Hi, > >> > >>I would welcome suggestions on how to lower the checkpoint durations at > >>one of our sites. Relevant ONCONFIG parameters are given below. All > >>chunks are cooked files. At checkpoint time, there are on the average > >>about 500 dirty buffers to flush. > > > >From what I have heard going to raw space might cut checkpoints down by 20% > > > Arrgh! Do not decrease LRU_MIN_DIRTY and LRU_MAX_DIRTY to 1 and 0! > > This will probably decrease %wcache in onstat -p output to < 85%. > > Instead may increase LRUS to 64 and CLEANERS to 32 > > We run with LRUS 127 and CLEANERS 48 and have reduced checkpoint > times from up to 20-120 seconds down to a 1-8 seconds. > > >> > >>Thanks. > >> > >>Alanoly Andrews. > >> > >>====================================== > >>Informix Dynamic Server 7.31 UC3 on AIX 4.3.2 > >> > >>----------------------------------- > >>Here are the last 20 lines of the Server Log: > >> > >>10:31:34 Checkpoint Completed: duration was 10 seconds. > >>10:36:46 Checkpoint Completed: duration was 11 seconds. > >>10:41:55 Checkpoint Completed: duration was 10 seconds. > >>10:47:07 Checkpoint Completed: duration was 11 seconds. > >>10:52:15 Checkpoint Completed: duration was 9 seconds. > >>10:57:25 Checkpoint Completed: duration was 9 seconds. > >>11:02:35 Checkpoint Completed: duration was 9 seconds. > >>11:05:16 Logical Log 161 Complete. > >>11:05:16 Logical Log 161 - Backup Started > >>11:05:19 Logical Log 161 - Backup Completed > >>11:07:46 Checkpoint Completed: duration was 11 seconds. > >>11:12:55 Checkpoint Completed: duration was 9 seconds. > >>11:18:04 Checkpoint Completed: duration was 9 seconds. > >>11:23:13 Checkpoint Completed: duration was 8 seconds. > >>11:28:25 Checkpoint Completed: duration was 11 seconds. > >>11:33:33 Checkpoint Completed: duration was 9 seconds. > >>11:38:53 Checkpoint Completed: duration was 19 seconds. > >>11:44:03 Checkpoint Completed: duration was 9 seconds. > >>11:49:10 Checkpoint Completed: duration was 8 seconds. > >>11:54:21 Checkpoint Completed: duration was 10 seconds.> >> > >>-------------------------------------- > >> > >>Relevant parts of the "onconfig" file: > >> > >>MULTIPROCESSOR 0 # 0 for single-processor, 1 for multi-> >>processor > >>NUMCPUVPS 1 # Number of user (cpu) vps > >>SINGLE_CPU_VP 1 # If non-zero, limit number of cpu vps> >>to one > >> > >>NOAGE 1 # Process aging > >>AFF_SPROC 0 # Affinity start processor > >>AFF_NPROCS 0 # Affinity number of processors> >> > >># Shared Memory Parameters > >> > >>LOCKS 150000 # Maximum number of locks > >>BUFFERS 45000 # Maximum number of shared buffers > >>NUMAIOVPS 2 # Number of IO vps> > > >I imagine this is to low. The advice usually given on this is as follows > >do onstat -g iov > >look at the io/wup column. If one of the aio > >has a value below 1 I imagine you have enough. > >Do not do this test till after increasing LRU's and CLEANERS. > > > >>PHYSBUFF 64 # Physical log buffer size (Kbytes) > >>LOGBUFF 64 # Logical log buffer size (Kbytes)> >>LOGSMAX 100 # Maximum number of logical log files > >>CLEANERS 2 # Number of buffer cleaner processes> > > >I recommend this be the same as LRU's. Cleaners flush your LRU queues. > > > >>SHMBASE 0x30000000 # Shared memory base address > >>SHMVIRTSIZE 100000 # initial virtual shared memory segment> >>size > >>SHMADD 64000 # Size of new shared memory segments > >>(Kbytes) > >>SHMTOTAL 720000 # Total shared memory (Kbytes). > >>0=>unlimited > >>CKPTINTVL 300 # Check point interval (in sec) > >>LRUS 6 # Number of LRU queues> > > >I would put this up at 50. This splits your buffers up into shorter LRU > >queus.. > > > >>LRU_MAX_DIRTY 10 # LRU percent dirty begin cleaning limit > >>LRU_MIN_DIRTY 5 # LRU percent dirty end cleaning limit> > > >If you only have 500 dirty buffers changing these wont help > > > >>LTXHWM 50 # Long transaction high water mark> >>percentage > >>LTXEHWM 60 # Long transaction high water mark > >>(exclusive) > >>TXTIMEOUT 0x12c # Transaction timeout (in sec) > >>STACKSIZE 64 # Stack size (Kbytes)> >> > > > >Hope these suggestions help. > > > >Will > > > >------------------------------------------------------------ > >This e-mail has been sent to you courtesy of OperaMail, as a free service > >from > >Opera Software, makers of the award-winning Web Browser, Opera. Visit us at > >http://www.opera.com/ or our portal at: http://www.myopera.com/ Your free e-mail > >account is waiting at: http://www.operamail.com/ > >------------------------------------------------------------ > > > > -- > David Williams