HW warning - Low Non-page pool
Hi, I'm still in developement phase working on operator workstation that will be used in plant. OS: Windows 7 Professional SP 1, 64-bit.
Every time I start workplace runtime, I get System message that my "Non-page pool is low (3xx,xMB used)".
As I found on some forums it is possible to change registry key at: [HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Session Manager\Memory Management], "PagedPoolSize"=dword:ffffffff for max pool size, but I want to keep my workstation stable, since windows support page says changing this can make it unstable. Is there another solution? And what is "support documentation" mentioned in message?
Voted best answer
I sincerely recommend to investigate the twitching further - please consider engaging your regional ABB support center in this process!
Hint: if you never used PowerShell here is a good appetizer:
Try these commands in a Terminal Window:
1)
powershell "get-process afwworkplaceapplication"
2)
powershell "get-process | select-object -name, npm, name | sort npm"
The first will list a number of process statistics for all workplaces in a node, NPM is one of them.
The second will return a sorted list on all processes' use of NPP.
In my lab the Plant Explorer and quad screen workplaces (in a 5.1 RevC x64 client) allocates a few hundred KB of NPP after start and opening of some process graphics, trends and some minimal engineering.
PS C:\Users\800xaadmin> get-process afwworkplaceapplication
Handles NPM(K) PM(K) WS(K) VM(M) CPU(s) Id ProcessName
------- ------ ----- ----- ----- ------ -- -----------
1756 188 252664 256336 775 15.76 1540 AfwWorkplaceApplication
2563 234 451380 384200 1009 255.47 5608 AfwWorkplaceApplication
The total use of NPP is reported as 87 MB and my top-10 NPP users looks like this:
35296 AfwFsdNode
36736 Batch_RedStatus
37432 AfwSMClient
38260 AbbClientLicenseProvider
38748 AfwAppLogService
42856 FrameworkService
46040 svchost
54704 svchost
65456 explorer
199140 AfwWorkplaceApplication
232972 AfwWorkplaceApplication
The warning limit is set on the Basic Computer Device aspect for each Basic Computer asset object and defaults to 150 MB. I believe the tooltip to be wrong (I think the total consumed level limit is set, not the remainder limit - but I'm not completely sure on this).
I assume you must press the Populate button in the IT Device Manager aspect to get a change in effect.
We can not forbit you from changing the default values, but doing so may result in loss of notification on loss/drain of NPP and consequential problems later on - so proceed with caution, favorably in cooperation with your local ABB support team.
/Stefan
Answers
I belive the actual NPP warning threshold used by PNSM is a preset value regardless of computer kind/role.
Have you checked what process(es) that drain NPP? Can you compare with another client/system?
I suggest that you contact your regional ABB support center and have them to verify your setup. If everything is OK the warning threshold for some nodes might be increased.
/Stefan
I got some WS time, here are the results:
1) 1st terminal command gives:
Handles NPM(K) PM(K) WS(K) VM(M) CPU(s) Id ProcessName
------- ------ ----- ----- ----- ------ -- -----------
1427 160 219636 246040 723 12.37 1112 AfwWorkplaceApplic...
430 41 22112 39556 228 1.87 2864 AfwWorkplaceApplic...
1234 142 152800 187272 633 7.69 8820 AfwWorkplaceApplic...
2) 2nd gives (top 10 users):
52112 explorer
53936 svchost
56140 AdvAeAlarmManager
58468 AfwSystemMessage
61564 AfwServiceManager
67140 AfwAppLogService
68584 AfwADServer
74456 explorer
144604 AfwWorkplaceApplication
163700 AfwWorkplaceApplication
Hope this gives some insight in problem.
3) I've tried to change warning limit to 350MB and now it doesn't activate anymore. Is it ok to leave problem at this, I mean, system will stay stable?
4) I haven't contact our ABB support yet, but will do if problem occurs again.
/ih
I can't find anything obviously wrong in the statistics you inserted.
The workplace and top-10 NPP usage is more or less the same when comparing with the figures from my lab.
I suggest comparing additional performance counters, process statistics, etc. between the problematic node and a node known to work correctly without NPP warnings.
Increasing the NPP warning threshold by more than a factor of 2x seems risky without knowing the underlying triggering factors.
I judge the PNSM NPP alarm + heavy twitching in CBM to be more than enough reasons to contact ABB support and have the system & configuration checked out.
/Stefan
Our system consists only of one node, so I can't compare. Thanks again, Stefan.
/ih
by ih Rank: 768 on 8/19/2013 11:45:35 PM | Like (0) | Report
Correction: "NonPagedPoolSize" for reg. key.
Add new comment