Re: AW: minfree maxfree maxpgahead
From: David Zarnoch (David.Zarnoch_at_BISYS.COM)
Date: 09/08/04
- Previous message: Anil Maurya: "Re: Quick question."
- Maybe in reply to: Uwe Gründel: "AW: minfree maxfree maxpgahead"
- Next in thread: Holger.VanKoll_at_SWISSCOM.COM: "Re: AW: minfree maxfree maxpgahead"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: Wed, 8 Sep 2004 08:12:32 -0500 To: aix-l@Princeton.EDU
If I use the calculation:
minfree=(120*Number of CPUs)
maxfree=(minfree+maxpgahaed)
This would give me:
minfree:
960=(120*8)
maxfree=(960 + maxpgahead)
Doesn't "960" seem kinda high?
Also, how do I calculate the value of "maxpgahead"?
Here's some additional info:
DiskName Busy Read Write
hdisk1 88% 161.4 1783.9KB
hdisk0 95% 181.1 1786.8KB
hdisk13 71% 171.7 321.4KB
There is a mirrored paging space on hdisk0 and hdisk1
Threr is an additional paging space on hdisk13
This "Busy" state remains pretty much constant all day.
Here are some memory stats from "nmon":
Memory Use Physical Virtual Paging pages/sec In Out VM parameters
% Used 100.0% 17.0% to Paging Space 82.4 71.0 numperm 52.1%
% Free 0.0% 83.0% to File System 560.1 216.7 minperm 9.6%
MB Used 40945.9MB 7079.3MB Page Scans 2463.9 maxperm 19.2%
MB Free 14.1MB 34584.7MB Page Cycles 0.0 minfree 504
Total(MB) 40960.0MB 41664.0MB Page Reclaim 0.0 maxfree 512
Here are some stats from vmstat:
kthr memory page faults cpu
----- ----------- ------------------------ ------------ -----------
r b avm fre re pi po fr sr cy in sy cs us sy id wa
4 6 6426746 3353 0 23 34 2729 3563 0 3137 38230 6339 22 12 42 24
2 3 6433062 9185 0 209 0 3467 16103 0 2872 18575 5203 24 7 23 47
3 2 6435917 1700 0 174 0 103 353 0 3107 31006 5333 23 12 17 48
2 4 6449838 1610 0 146 396 3441 15774 0 3023 17025 5017 19 8 19 55
1 2 6446988 1714 0 197 98 342 1281 0 2802 21137 4398 9 6 28 57
1 5 6449420 2660 0 215 439 1248 4738 0 2744 19123 4077 9 6 25 60
1 13 6460860 1278 0 133 1068 2655 8836 0 2976 15830 4194 6 8 15 71
1 9 6466418 3371 0 186 946 2215 7525 0 3270 22173 5180 12 9 14 66
1 15 6475654 3183 0 198 1045 2441 7814 0 2854 20926 4216 9 10 19 62
3 10 6451014 23806 0 341 215 808 2958 0 3551 21103 5876 13 9 21 57
1 7 6461095 9759 0 163 0 0 0 0 2646 21680 4229 9 13 25 53
2 7 6447999 18719 0 238 0 0 0 0 3136 24900 5261 11 14 16 59
1 3 6450260 12029 0 182 0 0 0 0 2833 17603 4513 7 6 32 55
1 6 6456301 1512 0 237 0 219 921 0 3101 22177 5075 10 8 26 57
1 3 6456947 1532 0 189 0 908 4913 0 3138 27693 5534 12 11 33 45
Thanks for all your help!
Dave Zarnoch
Bisys
>>> uwe.gruendel@ITCONCEPTS.DE 09/08/04 02:23AM >>>
Hi David
We tune it like the tuning-guide says and as you wrote:
minfree=(120*Number of CPUs)
maxfree=(minfree+maxpgahaed)
and it works fine.
But i don`t think, that this is your problem !
Please tell us a little bit more details.
Is your system paging ?
How is your memory used ?
Are you shure your performance bottleneck is memory ?
Uwe
-----Ursprüngliche Nachricht-----
Von: David Zarnoch [mailto:David.Zarnoch@BISYS.COM]
Gesendet: Dienstag, 7. September 2004 20:41
An: aix-l@Princeton.EDU
Betreff: minfree maxfree maxpgahead
Folks,
We have been having a lot of problems with a slow LPAR AIX5.2
I have been seeing a lot of articles that say:
"The difference between minfree and maxfree should always be equal to or greater than maxpgahead. "
Here are our values (vmtune)
maxfree = 512
minfree = 504
maxpgahead = 256
This tells me that 8 < 256
These values seem to be defaults
Is the above statement correct or are parameters WAY off base?
Thanks!
Dave Zarnoch
Bisys
- Previous message: Anil Maurya: "Re: Quick question."
- Maybe in reply to: Uwe Gründel: "AW: minfree maxfree maxpgahead"
- Next in thread: Holger.VanKoll_at_SWISSCOM.COM: "Re: AW: minfree maxfree maxpgahead"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Relevant Pages
|
|