question

Sebastian K. avatar image
Sebastian K. asked ·

[Appmon agent] - After converting IIS Classic Agents into Appmon Agents page faults per secon increased

Hello,

did anybody has such problem? After migration appmon 7.0 + Classic agent into appmon 7.2.1 Appmon agent on IIS servers I see that page faults per second increased from 100-200 to 1000-2500. Memory looks ok on those hosts (nothing has changed from time before migration). What can be reason of such behavior?

Regards,
Sebastian

appmonagentappmon 77.2iis
10 |2000000 characters needed characters left characters exceeded

Up to 10 attachments (including images) can be used with a maximum of 50.0 MiB each and 250.0 MiB total.

Sebastian K. avatar image
Sebastian K. answered ·

Yes, you have to disable the collection of ETW metrics by setting the agent debug flag debugEnableEtwProfilingNative to false on Agent Group. This mechanism is creating and updating multiple small files. When you have multiple application pools on single host it causes page faults and iops issues.

Sebastian

7 comments Share
10 |2000000 characters needed characters left characters exceeded

Up to 10 attachments (including images) can be used with a maximum of 50.0 MiB each and 250.0 MiB total.

Hello @sebastian k.

How can I add the debug flag on the agent group instead of a mapping agent?

Do I need to restart the agents/processes to take effect after adding the debug flag?

Regards,

Babar

0 Likes 0 · ·

You are adding flag to agent group in debug mode (ctrl + shift + f9). Restart is not needed

0 Likes 0 · ·

Hello @sebastian k.

Thank you for the confirmation about restart.

In agent group we have number of unique agents name for the mapping, therefore, I wanted to know a way to apply the debug in one shot instead of individual agent.

Regards,

Babar

0 Likes 0 · ·

Ctrl shift f9 to go into debug mode. Pick agent group -> debug options and there you can add flag to whole agent group

0 Likes 0 · ·

Hello @sebastian k.

I guess we cannot pick the agent group instead the agent only to apply the debug flag.

Can you send me a screeshot?

Regards,

Babar

0 Likes 0 · ·

Here you have path to add debug flag to Agent Group. I don't know how you want to add it to single agent actually.

Sebastian

0 Likes 0 · ·

Hello @sebastian k.

To be very honest I am very well aware about debug settings.

For the agents mapping we have different strategies, therefore, we create unique name of each agent to map.

I tired and found there is no way to apply a debug flag at once on all alias/name in the agent group.

Regards,

Babar

0 Likes 0 · ·
debug-flag.jpg (13.4 KiB)
Bent K. avatar image
Bent K. answered ·

We had the same issue. Also the CPU usage went up form 30% to 90%, and I saw that ETW was causing it with high I/O. Since we have more than 10 applications w3wp-processes per host, I guess the impact is higher than average. We had to set the debugEnableEtwProfilingNative to off.

It worries me that this huge performance hit is default behaviour on Agent Platfrom Switch.

1 comment Share
10 |2000000 characters needed characters left characters exceeded

Up to 10 attachments (including images) can be used with a maximum of 50.0 MiB each and 250.0 MiB total.

It is not common to have such amount of application pools on single host. But I agree there should be worning somwhere in appmon

0 Likes 0 · ·
Joe H. avatar image
Joe H. answered ·

This indeed seems strange. I'm wondering if it's a units issue with how it's being displayed. Can you clarify how you're measuring page faults?

Bottom line, i'd suggest you reach out to support and have them dig into this a bit deeper.

2 comments Share
10 |2000000 characters needed characters left characters exceeded

Up to 10 attachments (including images) can be used with a maximum of 50.0 MiB each and 250.0 MiB total.

Support ticket has been opend. Page faults are measured by custom chart and on host dashboard as well. Performance wherehouse was disconnected for 3 days, support suggest to wait few days because of possible error with data presentation.

If situation will be permanent we will use perfview and compare results.

Regards,
Sebastian

1 Like 1 · ·

Hello @sebastian k.

We are facing the same issue with the IIS/.NET after switching the platform from Classic to AppMon.

For the time being I increased the maximum page faults threshold from 30 to 500 just to avoid the fault positive.

Did you resolve the issue?

Regards,

Babar

0 Likes 0 · ·