Scheduler patch 03 available for testing
Matthew Dillon
dillon at apollo.backplane.com
Sat Jul 24 13:35:13 PDT 2004
:However, it is now easy to notice that after having been relieved from
:heavy loads the system appears to treat some of the remaining processes
:as if the system's primary resources are still scarce. The result is
:some applications remain less responsive, long after they should have
:returned to being "snappy". I suppose this could have something to do
:with swapping pages back into primary memory after the system load has
:been lightened.
If your system is paging a lot, then the data related to these processes
would have to be paged in again.
Running a 'vmstat 1' or a 'systat -vm 1' in an xterm while playing
around and observing the results may allow you to characterize the
problem a bit better.
-Matt
Matthew Dillon
<dillon at xxxxxxxxxxxxx>
More information about the Kernel
mailing list