RE: Onstat -P output

Hi Leona. If anything I say in here evokes a response 'I know that dummy'(-:)) I apologize. Given your profile, Onconfig and the fact that you are running an OLTP I would suggest you make the following changes to your onconfig. Some are calculations, some out of experience: You profile gives tells you your instance uses 3063808 Kbytes of memory. You initial SHMVIRTSZ is set to 294912 Kbytes. Now, when started up the instance will simply take what it needs base on buffers, poll threads, ds,locks etc, etc, etc so you can assume the number in the profile is what is takes when started. Now, when you set SHMVIRTSZ to 3063808 Kbytes you might end up with a problem because the instance will usually double that number. I suggest you set it to 1914880KB (that is half it uses according to your profile times 1.25 to allow for a little buffer without having to assign new shared memory segments.) When started up onstat - will probably show double that (approx 3900000KB) or a bit more. Whenever you get the possibility, rebuild your instance and have the root chunk be used for nothing else but the instance tables, so do not create a database in it (at all). I can't check how much your root dbs has filled, so it is just a remark, but not enough space for your instance will result in tremendous performance loss and the instance not working at all. I case you haven't done this make sure your logical logs are in a dbspace of there own and not in your root. Whenever possible do the same with your physical log. Put it in a chunk of its own, not sharing with anything else and no offset either. If you loose space don't worry about it. Limit SHMTOTAL to 60%of what you have (8GB*.6)= SHMTOTAL 5033164 which will leave the instance with plenty of room and makes sure your system will not run into trouble (too soon). When the instance starts you may find a 'SHMTOTAL rounded to .....' in the online.log. When that happens just adjust the number in the onconfig to that. Change: MAX_PDQ_PRIORITY 60 (instead of 100) to prevent dss bringing the instance to a virtual halt) DS_TOTAL_MEMORY 1258290KB ( 25% of SHMTOTAL )DS_MAX_QUERIES (empty, so the instance calculates max queries) After restarting your engine take a look in the online log and note the number the instance assigns to D_MAX_QUERIES. With that number recalculate your RA_ parameters. Your RA_pages seems way to high for an OLTP and probably causing a lot of useless pages in your buffers. RA_PAGES (nbr_of_buffers * use% of buffers) / ( 2* [nbr of max queries]) + 2 Say 1200000 * .25 [for 25%] ) / ( 2* [ say 14000 =28000] ) + 2 RA_THRESHOLD (nbr_of_buffers * use% of buffers) / ( 2* [nbr of max queries]) - 2 Say 1200000 * .25 [for 25%] ) / ( 2* [ say 14000 =28000] ) - 2 You will probably end up with something like 28 and 24 TRY and set optcompind to 0 I do not know how many discs there are in you setup (even though you are using raid), but you may want to take a look at your LRU and CLEANERS as well. Just in case it is not yet the case: Have you OS administrators check the way your raid is set up and how they distributed the discs. You would want your instance spread over as many raid controllers as possible. Split your instance over more than 1 volume group, but keep them balanced for i/o When in need 'just holler' Regards, ------------------------------------------------------------------------ R.A. Reissaus Risdi, Zurich, Switzerland Ibm/Informix Consultants Telefoon: +41(0)43 534 67 94 : Phone Mobiel : +41(0)79 592 99 13 : Mobile E-mail : r.a.reissaus@risdi.com Website : www.risdi.com ------------------------------------------------------------------------ Disclaimer: `Externe E-Mail wordt door partijen niet gebruikt voor het aangaan van verplichtingen` `Any e-mail messages from sending parties named in this e-mail are given in good faith but shall not be binding nor shall they be construed as constituting any obligation on the part of any such party.` ------------------------------------------------------------------------ -----Oorspronkelijk bericht----- Van: owner-informix-list@iiug.org [mailto:owner-informix-list@iiug.org] Namens Leona Ankrah Verzonden: donderdag 18 augustus 2005 16:52 Aan: informix-list@iiug.org Onderwerp: Onstat -P output Hi, I'm using IDS9.21 UC6 running on a SOlaris 8.0 Operating system. The output of the onstat -P is not acceptable and I would appreciate it if I could be helped get a better results. The system is not performing well enough. Attached is the output of the onstat -p. I have also attached the onconfig file for the database configuration. please help. Leona Ghana Telecom Limited __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com sending to informix-list