swapinfo

[Michael_425]

Hi all

I have an n 9000 with 8Gb Physical memory, glance is showing mem usage at %100 and swap info at %55. The output from swapinfo is as follows...
Kb Kb Kb PCT START/ Kb
TYPE AVAIL USED FREE USED LIMIT RESERVE PRI NAME
dev 1982464 1360472 621992 69% 0 - 1 /dev/vg00/lvol2
localfs 4096000 0 4096000 0% 4096000 0 1 /x/paging
localfs 3072000 0 3072000 0% 3072000 0 1 /d/paging
reserve - 5628132 -5628132
memory 6713576 1626664 5086912 24%

Should I be concerned about this?

Thanks in advance

[Michael_425]

Sorry about that, I should probably have included this output instead... from swapinfo -tam

Mb Mb Mb PCT START/ Mb
TYPE AVAIL USED FREE USED LIMIT RESERVE PRI NAME
dev 1936 1314 622 68% 0 - 1 /dev/vg00/lvol2
localfs 4000 0 4000 0% 4000 0 1 /x/paging
localfs 3000 0 3000 0% 3000 0 1 /d/paging
reserve - 5443 -5443
memory 6556 1588 4968 24%
total 15492 8345 7147 54% - 0 -

[Bill Hassell · Accepted solution]

Your primary swap area is 70% utilized but you have a lot of extra swap space available in the local filesystems. So you won't run out of virtual memory. However, if you have performance problems (things are running slower than expected), then the swap usage may be a concern. It means that there is not enough memory for all the processes and data. This will only be a problem if paging (swapping) is occurring all the time. Use vmstat and look at the po (page-out) column. Single digits are OK, double digits are a warning, and more than 2 digits means a serious lack of RAM exists. Of course these numbers would be a concern if they are steady. It is OK for a burst of paging to occur and then be zero for most of the time.

There are only 2 fixes for excessive paging (swap usage):

1) reduce the number of programs running at the same time, or ask the authors of the programs to redecue the amount of RAM they are using. Reducing RAM usage in a database almost always means slower response from the database.

2) Buy more RAM.