... at the level that the SQL engine that throws the error runs at, swap and physical memory are indistinguishable. For example try just filling up a string in memory with junk. When you run out swap space, your app will throw an out of memory error because the OS cannot allocate any more RAM to it, but the IS will start tossing low swap space warnings and maybe try to increase swap if possible.
Now, that being said, the real question is, "why is the database suddenly using a heck of a lot more RAM (physical + swap) in the first place?" Increasing swap may very well just put off a much bigger problem. You might want to look into why the DB now needs a lot more RAM than it used to.
J.Ja
Discussion on:
View:
Show:
I haven't asked the DBA to look but I would be willing to bet it's just the Portal using up all that RAM and swap space. As we've rolled out the Portal to more and more users, there's now gigabytes worth of document blobs stored in that schema. Plus many more users logged in, so there's session and state objects being maintained by the Portal. Etc., etc.
we are experiencing the same problem at the moment with a windows 2003 enterprise server with 8 gig of ram. Have you tuned the OS.
have a look at these if you are using windows that is:
http://www.dba-oracle.com/oracle_tips_ram_waste.htm
http://dba.ipbhost.com/index.php?showtopic=4437
have a look at these if you are using windows that is:
http://www.dba-oracle.com/oracle_tips_ram_waste.htm
http://dba.ipbhost.com/index.php?showtopic=4437
Hey, thanks for the pointers. Our configuration is pretty similar to yours -- 8 GB Windows 2003.
The title is simplistic, but James has it right. The system uses RAM to hold data. When RAM is full, and on more efficient systems when RAM isn't getting used, it is swapped to disk if there is space available.
How did you analyze your RAM usage? Tools like 'top' and 'sar' on a Unix box at the time of the error should have showed a lot of disk activity (swapping) and 0 available RAM.
If your database appears to have a lot of RAM available then you might try reducing some of your init parameters. It maybe that the database has enough RAM, but that the OS is starved.
Couldn't do more analysis without the details.
Might be useful to see how you troubleshot the problem if you do another article.
How did you analyze your RAM usage? Tools like 'top' and 'sar' on a Unix box at the time of the error should have showed a lot of disk activity (swapping) and 0 available RAM.
If your database appears to have a lot of RAM available then you might try reducing some of your init parameters. It maybe that the database has enough RAM, but that the OS is starved.
Couldn't do more analysis without the details.
Might be useful to see how you troubleshot the problem if you do another article.
- Keyboard Shortcuts:
- Prev
- Next
- Toggle
