Hi all!
Working on Oracle v8i/9i on Unix Sun Solaris v8.0.
I am experiencing excessive paging & Swapping.Would like to know the cause.
I guess:could be due to inappropriate setting of Unix Kernel Parameters...
Please correct me if I am wrong!
Thanks&Regards,
Amit.
How much memory does the system have? How much swap space? Has the /etc/system file been updated with the required parameters for Oracle? How long has the problem been occuring? (and what changed when it started - more users, more databases, new searches being launched daily...)
This isn't something one of us can just say "Yea, you need more memory" You have to work with your Oracle DBA and IT team to find out what was happening and what is happening, know/investigate any changes happening on the system be it OS or Oracle.
Give more details of what you have found and possibly a 'best guess' can be given back.
Post output from any of the monitoring tools showing your problem. If you haven't been watching your servers or at least have sar running and collecting information, turn it on to have some clue of what the server is doing at what times. Check for new cron jobs, increase in users utilizing the database, changes in your volume management (if you are using Vertias or any other volume manager, losing a disk drive and having it 'automatically' place databases on different disks that had the space (but really could not handle the increased load of hits) could cause problems (if you didn't have hot spares).
Hi RTM,
Thanks you for your time.
My Guess is that SGA could be larger in size or too much of memory.
May be the Unix Kernel Parameters are not set appropriately like LOTSFREE,DESFREE, MINFREE which are Paging & Swapping Parameters.
Any idea!May be checking on these could give me the clue.
Thansk & regards,
Amit.
Please reread RTM's post. He asked that you post the answers to several questions. We really need that information.
Also use our search function. Search for paging and swapping. You will get a lot of threads that discuss this.
It really isn't as complicated as people are making it.
If your experiencing excessive paging and swapping like you said, get a verification report from the output of commands like iostat, vmstat, sar -wpgr 5, ps and then send it to the person in charge of adding memory to the system. Have him study the report and determine if a memory upgrade is needed.
If you find out that the excessive swapping or whatever problem your having is because of memory shortage, you might wanna try a few things (for the main time) before deciding to buy and install new memory. I suggest you take a good look at ps and find out the processes that take up a lot of memory or processes that accumulate a lot of time. Contact the Users who owns this processes and ask if you can reschedule the processes to run at a later time when the load on the system is light. Or you can ask the user if you can renice the job to run at a lower priority.
Thats all there is to it, my man.
Amit
The default sizes of the rollback segments and redo logs are very small. Do you have all these tablespaces on single disk. Use system tools like iostat etc. and locate which is causing the excessive IO. Also you can use oracle tracing and tkprof tool to estimate the statistics of the database operations. This will give you the details of CPU time, disk usage, indexing issues if any