From: fmu@OERAG.DE
Date: Thu Nov 14 2002 - 05:59:38 EST
Hm,
I think your definition from num-, max-, minfree are right (I think
so, but I'm not really sure)
If the reduction of the values maxperm/minperm a success does not have,
possibly the application is too large for your memory. Check your box with
svmon....
For example:
svmon -Uut 10
svmon -Put 10
and check the memory which your application load in the memory. Add the
values for 'EXCLUSIVE Segments'. Is ist more than 6GB?
I'm not a specialist in memory management, but I think so ;-)
Mit freundlichen Grüßen / With best regards,
Frank Müller
-------------------------------------------------------
| OERAG Rechtsschutzversicherungs AG |
| |
| Dipl.-Physik Ing. Frank Mueller |
| IBM Certified Specialist - AIX |
| Fachverantwortlicher UNIX Systems |
| OERAG IT , Rechenzentrum |
| |
-------------------------------------------------------
oracle runs here, several instances.
there is much io on sequential files > 1GB, aprox. 15mbyte/sec. box has
6gb,6cpu. so I dont gain much from filecache.
I can set maxperm even lower, still have paging:
number of valid memory pages = 1572825 maxperm=4.0% of real memory
maximum pinable=80.0% of real memory minperm=2.0% of real memory
number of file memory pages = 228461 numperm=14.5% of real memory
Is the following correct:
If numperm > maxperm, only filepages will be stolen. So there should not be
paging-space usage.
If minperm > numperm > maxperm filepages will be stolen if their
repage-rate is higher than those of working storage.
The problem is long responsetime to short queries. So I set minfree to 600
and the responsetime increased a bit.
Anyway, I just dont understand why paging-space is used as long there are
more filepages then maxperm.
This archive was generated by hypermail 2.1.7 : Wed Apr 09 2008 - 22:16:20 EDT