SPINCNT configuration parameter in 5.10

Informix Version 5.10 Hardware: HP-K 4-way that was upgraded from a single cpu. OS: HP-UX 10.20 The application runs on the server and uses shared memory connections. I am working on cpu bottlenecks and I found the following. According to the book 'Informix Performance Tuning S.E' by Elizabeth Suto SPINCNT should be 0 on single cpu and the default is 300. Currently this config parameter is set to 0 and I'm assuming this is because the server was upgraded from a single to 4-way. What advice or experience can anyone give me on tuning this parameter? Are there other things that I should be looking for to reduce cpu bottlenecks? Best regards, -- Laurence

L H wrote: > Informix Version 5.10 > Hardware: HP-K 4-way that was upgraded from a single cpu. > OS: HP-UX 10.20 > > The application runs on the server and uses shared memory connections. I > am working on cpu bottlenecks and I found the following. > > According to the book 'Informix Performance Tuning S.E' by Elizabeth Suto > SPINCNT should be 0 on single cpu and the default is 300. > > Currently this config parameter is set to 0 and I'm assuming this is > because the server was upgraded from a single to 4-way. > > What advice or experience can anyone give me on tuning this parameter? Reset it to the default? > Are there other things that I should be looking for to reduce cpu > bottlenecks? On what basis do you conclude you have CPU bottlenecks? -- Ciao, The Obnoxious One "Ogni uomo mi guarda come se fossi una testa di cazzo"

"Obnoxio The Clown" <obnoxio@hotmail.com> wrote in message news:blnfj7$ef7j6$1@ID-64669.news.uni-berlin.de... > L H wrote: > > > Informix Version 5.10 > > Hardware: HP-K 4-way that was upgraded from a single cpu. > > OS: HP-UX 10.20 > > > > The application runs on the server and uses shared memory connections. I > > am working on cpu bottlenecks and I found the following. > > > > According to the book 'Informix Performance Tuning S.E' by Elizabeth Suto > > SPINCNT should be 0 on single cpu and the default is 300. > > > > Currently this config parameter is set to 0 and I'm assuming this is > > because the server was upgraded from a single to 4-way. > > > > What advice or experience can anyone give me on tuning this parameter? > > Reset it to the default? > > > Are there other things that I should be looking for to reduce cpu > > bottlenecks? > > On what basis do you conclude you have CPU bottlenecks? At first I did not think there was a cpu bottleneck. The sar data was showing disk and memory problems. I added memory and now I'm changing the disk configuration to balance the I/O for the filesystem. The disk layout for the database was done well, but the FS layout had issues. I recently installed Glance and immediately saw the cpu alarms go off. It showed the idle times at almost zero and the que's are backing up. The Glance advisor keeps showing cpu bottlenecks. > > -- > Ciao, > The Obnoxious One > > "Ogni uomo mi guarda come se fossi una testa di cazzo" Laurence

L H wrote: > > "Obnoxio The Clown" <obnoxio@hotmail.com> wrote in message > news:blnfj7$ef7j6$1@ID-64669.news.uni-berlin.de... >> L H wrote: >> >> > Informix Version 5.10 >> > Hardware: HP-K 4-way that was upgraded from a single cpu. >> > OS: HP-UX 10.20 >> > >> > The application runs on the server and uses shared memory connections. >> > I am working on cpu bottlenecks and I found the following. >> > >> > According to the book 'Informix Performance Tuning S.E' by Elizabeth > Suto >> > SPINCNT should be 0 on single cpu and the default is 300. >> > >> > Currently this config parameter is set to 0 and I'm assuming this is >> > because the server was upgraded from a single to 4-way. >> > >> > What advice or experience can anyone give me on tuning this parameter? >> >> Reset it to the default? >> >> > Are there other things that I should be looking for to reduce cpu >> > bottlenecks? >> >> On what basis do you conclude you have CPU bottlenecks? > > At first I did not think there was a cpu bottleneck. The sar data was > showing disk and memory problems. I added memory and now I'm changing the > disk configuration to balance the I/O for the filesystem. The disk layout > for the database was done well, but the FS layout had issues. I recently > installed Glance and immediately saw the cpu alarms go off. It showed the > idle times at almost zero and the que's are backing up. The Glance advisor > keeps showing cpu bottlenecks. Can you post your tbconfig file? How many shared memory chunks does ipcs report for the engine? -- Ciao, The Obnoxious One "Ogni uomo mi guarda come se fossi una testa di cazzo"

"Obnoxio The Clown" <obnoxio@hotmail.com> wrote in message news:blolk6$ek7fm$1@ID-64669.news.uni-berlin.de... > L H wrote: > > > > > "Obnoxio The Clown" <obnoxio@hotmail.com> wrote in message > > news:blnfj7$ef7j6$1@ID-64669.news.uni-berlin.de... > >> L H wrote: > >> > >> > Informix Version 5.10 > >> > Hardware: HP-K 4-way that was upgraded from a single cpu. > >> > OS: HP-UX 10.20 > >> > > >> > The application runs on the server and uses shared memory connections. > >> > I am working on cpu bottlenecks and I found the following. > >> > > >> > According to the book 'Informix Performance Tuning S.E' by Elizabeth > > Suto > >> > SPINCNT should be 0 on single cpu and the default is 300. > >> > > >> > Currently this config parameter is set to 0 and I'm assuming this is > >> > because the server was upgraded from a single to 4-way. > >> > > >> > What advice or experience can anyone give me on tuning this parameter? > >> > >> Reset it to the default? > >> > >> > Are there other things that I should be looking for to reduce cpu > >> > bottlenecks? > >> > >> On what basis do you conclude you have CPU bottlenecks? > > > > At first I did not think there was a cpu bottleneck. The sar data was > > showing disk and memory problems. I added memory and now I'm changing the > > disk configuration to balance the I/O for the filesystem. The disk layout > > for the database was done well, but the FS layout had issues. I recently > > installed Glance and immediately saw the cpu alarms go off. It showed the > > idle times at almost zero and the que's are backing up. The Glance advisor > > keeps showing cpu bottlenecks. > > Can you post your tbconfig file? How many shared memory chunks does ipcs > report for the engine? > > -- > Ciao, > The Obnoxious One > > "Ogni uomo mi guarda come se fossi una testa di cazzo" Here's the config and some performance data. Configuration File: /usr/informix/etc/tbconfig #************************************************************************** # # INFORMIX SOFTWARE, INC. # # Title: tbconfig # Sccsid: %W% %G% %U% # Description: INFORMIX-OnLine Configuration Parameters # #************************************************************************** # Root Dbspace Configuration ROOTNAME rootdbs # Root dbspace nameROOTPATH /usr/informix/links/rootdbs_chunk1 ROOTOFFSET 0 # Offset of root dbspace into device (Kbytes) ROOTSIZE 100000 # Size of root dbspace (Kbytes) # Disk Mirroring Configuration MIRROR 0 # Mirroring flag (Yes = 1, No = 0) MIRRORPATH # Path for device containing root dbspacemirror MIRROROFFSET 0 # Offset into mirror device (Kbytes) # Physical Log Configuration PHYSDBS logdbs # Name of dbspace that contains physical log PHYSFILE 50000 # Physical log file size (Kbytes) # Logical Log Configuration LOGFILES 15 # Number of logical log files LOGSIZE 10000 # Size of each logical log file (Kbytes) # Message Files MSGPATH /usr/informix/online.log # OnLine msg log pathname CONSOLE /usr/informix/online.log # System console msg pathname # Archive Tape Device TAPEDEV /dev/rmt/DLT_DRIVE TAPEBLK 32 # Archive tape block size (Kbytes) TAPESIZE 4000000 # Max amount of data to put on tape (Kbytes) # Logical Log Backup Tape Device LTAPEDEV /dev/null LTAPEBLK 20 # Logical log tape block size (Kbytes) LTAPESIZE 2000000 # Max amount of data to put on log tape (Kbytes) # Identification Parameters SERVERNUM 2 # Unique id associated with this OnLineinstance DBSERVERNAME xxxxxxxxx # Unique name of this OnLine instance # Shared Memory Parameters RESIDENT 0 # Forced residency flag (Yes = 1, No = 0)USERS 500 # Maximum number of concurrent user processes TRANSACTIONS 500 # Maximum number of concurrent transactions LOCKS 256000 # Maximum number of locks BUFFERS 60000 # Maximum number of shared memorybuffers TBLSPACES 8000 # Maximum number of active tblspaces CHUNKS 16 # Maximum number of chunks DBSPACES 8 # Maximum number of dbspaces and blobspaces PHYSBUFF 128 # Size of physical log buffers (Kbytes) LOGBUFF 32 # Size of logical log buffers (Kbytes)LOGSMAX 60 # Maximum number of logical log files CLEANERS 5 # Number of page-cleaner processes SHMBASE 0x0 # Shared memory base address CKPTINTVL 480 # Checkpoint interval (in seconds) LRUS 32 # Number of LRU queues LRU_MAX_DIRTY 10 # LRU modified begin-cleaning limit (percent) LRU_MIN_DIRTY 2 # LRU modified end-cleaning limit (percent) LTXHWM 50 # Long TX high-water mark (percent) LTXEHWM 60 # Long TX exclusive high-water mark (percent) # Machine- and Product-Specific Parameters DYNSHMSZ 0 # Dynamic shared memory size (Kbytes) GTRID_CMP_SZ 32 # Number of bytes to use in GTRID comparision DEADLOCK_TIMEOUT 60 # Max time to wait for lock in distributed env. TXTIMEOUT 300 # Transaction timeout for I-STAR (in seconds)SPINCNT 0 # No. of times process tries for latch STAGEBLOB # Reserved for INFORMIX-OnLine/Optical # System Page Size BUFFSIZE 2048 # Page size (do not change!) INFORMIX STATUS:COMPUTER NAME: xxxxxxx TIME: Sun Oct 5 00:00:01 PDT 2003 RSAM Version 5.06.UC2 -- On-Line -- Up 4 days 18:24:30 -- 153696 Kbytes CPU UTILIZATION: 12:00am up 4 days, 18:26, 3 users, load average: 0.26, 0.19, 0.18 RATIOS: READ CACHE WRITE CACHE LOCK HIT% LATCH HIT% USER/SYSTEM 98.28 84.76 100 100 5-1 WAITS: CHECKPOINTS BUFFERS DEADLOCKS 4306 1303039 67 LOGS: There are now 1 log(s) in use. This includes the current log: c18caeb8 6 U---C-L 125844 409478 5000 389 7.78 CHECKPOINTS: The last six checkpoints occurred as follows: 23:17:37 23:26:09 23:34:08 23:42:09 23:50:41 23:58:40 =========================================================================== CONTINUOUS BACKUP OF LOG FILES: OFF ARCHIVE: OFF rootdbs is 2.04 percent full and has 48978 free pages datadbs1 is 63.44 percent full and has 182799 free pages datadbs2 is

L H wrote: > "Obnoxio The Clown" <obnoxio@hotmail.com> wrote in message > news:blolk6$ek7fm$1@ID-64669.news.uni-berlin.de... >> L H wrote: >> >> > Informix Version 5.10 Yet, he also wrote: > RSAM Version 5.06.UC2 -- On-Line -- Up 4 days 18:24:30 -- 153696 Kbytes Shome mishtake, shurely? And then he also said >> >> > Are there other things that I should be looking for to reduce cpu >> >> > bottlenecks? >> >> >> >> On what basis do you conclude you have CPU bottlenecks? >> > >> > At first I did not think there was a cpu bottleneck. The sar data was >> > showing disk and memory problems. I added memory and now I'm changing > the >> > disk configuration to balance the I/O for the filesystem. The disk > layout >> > for the database was done well, but the FS layout had issues. I >> > recently installed Glance and immediately saw the cpu alarms go off. It >> > showed > the >> > idle times at almost zero and the que's are backing up. The Glance > advisor >> > keeps showing cpu bottlenecks. > RSAM Version 5.06.UC2 -- On-Line -- Up 4 days 18:24:31 -- 153696 Kbytes > > ovtbls ovlock ovuser ovbuff usercpu syscpu numckpts flushes > 0 0 0 0 89501.86 17707.74 694 1388 Gee, it's amazing how you get used to the IDS monitoring capabilities. Anyway, but resorting to some pretty creaky math, it looks like you've used 30 hours of CPU time in 114 hours, not exactly flat out, but it may well be that you have some critical load where the engine gets busy and the application is also using the machine. Or does the application reside elsewhere? > I'm concerned about the ckpwaits ratio and the number of deadlocks. > Any comments on this? Deadlocks are an application problem. Take you programmers outside, feed them on fine wine and foie gras for a month, then garotte them. Afterwards, you can have their livers with a nice chianti and some fava beans. Although I prefer Barolo myself. -- Ciao, The Obnoxious One "Ogni uomo mi guarda come se fossi una testa di cazzo"

"Obnoxio The Clown" <obnoxio@hotmail.com> wrote in message news:blq2dj$egpk0$1@ID-64669.news.uni-berlin.de... > L H wrote: > > > "Obnoxio The Clown" <obnoxio@hotmail.com> wrote in message > > news:blolk6$ek7fm$1@ID-64669.news.uni-berlin.de... > >> L H wrote: > > >> >> > Informix Version 5.10 > > Yet, he also wrote: Yes my mistake. I have another server with 5.10 and it seems that I'm a bit confused. > > > RSAM Version 5.06.UC2 -- On-Line -- Up 4 days 18:24:30 -- 153696 Kbytes > > Shome mishtake, shurely? > > > RSAM Version 5.06.UC2 -- On-Line -- Up 4 days 18:24:31 -- 153696 Kbytes > > > > ovtbls ovlock ovuser ovbuff usercpu syscpu numckpts flushes > > 0 0 0 0 89501.86 17707.74 694 1388 > > Gee, it's amazing how you get used to the IDS monitoring capabilities. Yes, IDS monitoring makes me a spoiled child too. > Anyway, but resorting to some pretty creaky math, it looks like you've used > 30 hours of CPU time in 114 hours, not exactly flat out, but it may well be > that you have some critical load where the engine gets busy and the > application is also using the machine. Or does the application reside > elsewhere? As I stated in my original post, the app runs on the server. The big loading occurs when the users are active. > > > I'm concerned about the ckpwaits ratio and the number of deadlocks. > > Any comments on this? Any comments on the ckpwaits ratio? > > Deadlocks are an application problem. Take you programmers outside, feed > them on fine wine and foie gras for a month, then garotte them. Afterwards, > you can have their livers with a nice chianti and some fava beans. > > Although I prefer Barolo myself. Ah yes, the brilliant and psychotic serial gormet treatment! > > -- > Ciao, > The Obnoxious One > > "Ogni uomo mi guarda come se fossi una testa di cazzo" Regards, Laurence

L H wrote: > "Obnoxio The Clown" <obnoxio@hotmail.com> wrote in message > news:blq2dj$egpk0$1@ID-64669.news.uni-berlin.de... >> L H wrote: >> >> > "Obnoxio The Clown" <obnoxio@hotmail.com> wrote in message >> > news:blolk6$ek7fm$1@ID-64669.news.uni-berlin.de... >> >> L H wrote: >> >> >> >> > Informix Version 5.10 >> >> Yet, he also wrote: > > Yes my mistake. I have another server with 5.10 and it seems that I'm a > bit confused. > >> >> > RSAM Version 5.06.UC2 -- On-Line -- Up 4 days 18:24:30 -- 153696 > Kbytes >> >> Shome mishtake, shurely? >> >> > RSAM Version 5.06.UC2 -- On-Line -- Up 4 days 18:24:31 -- 153696 > Kbytes >> > >> > ovtbls ovlock ovuser ovbuff usercpu syscpu numckpts flushes >> > 0 0 0 0 89501.86 17707.74 694 1388 >> >> Gee, it's amazing how you get used to the IDS monitoring capabilities. > > Yes, IDS monitoring makes me a spoiled child too. > >> Anyway, but resorting to some pretty creaky math, it looks like you've > used >> 30 hours of CPU time in 114 hours, not exactly flat out, but it may well > be >> that you have some critical load where the engine gets busy and the >> application is also using the machine. Or does the application reside >> elsewhere? > > As I stated in my original post, the app runs on the server. > The big loading occurs when the users are active. Maybe it's the application code and not the database? >> > I'm concerned about the ckpwaits ratio and the number of deadlocks. >> > Any comments on this? > > Any comments on the ckpwaits ratio? I wouldn't worry about the Checkpoint Waits. -- Ciao, The Obnoxious One "Ogni uomo mi guarda come se fossi una testa di cazzo"