I am currently recieving warning alerts from the Mendix Cloud, that my application has a critical CPU load e.g. WARNING - load average: 0.40, 0.51, 0.97. Then 5 minutes later I recieve an email that it has recovered again. To try and find out what the problem is I first checked the running processes and log to see if something large microflow was running. However, there was nothing special happening. When I open the monitoring trends page I also look at the graph for Application and Database node CPU usage. I do not see a load as mentioned in the warning email. The CPU usage is shown from 0 to 400%. I presume this is because the virtual machine has 4 processors. In the graph It does not seem to exceed the 60%. The only thing that is out of the ordinary is the JVM thread pool which seems to have become 4 times larger than the last time. We have upgraded our project from 5.16.1 to 5.20 along with some minor changes to the model. Can anyone help guess what the problem may be?
The warnings on the load seem to come from an issue with the monitoring setup.
I've seen this before, the app has multiple processors let's say 4 then the max load would be 4.0.
The monitoring however still thinks the app has 1 processor so a load of 0.97 will trigger the notification.
Correlating the actions to the CPU load is not an easy task and does not really seem necessary with the current load. The max load on the app is 0.97/4.0 *100 = 24.25%.
I would file a ticket with support to fix the notification settings.
Erwin 't Hoen
When we upgraded our app from Mendix 5.16.1 to 5.19.0 we experienced exactly the same:
JVM threads suddenly went from 200 to 500
CPU usage went went up a bit, but is nowhere near worrying values
Load average went through the roof: before upgrading it averaged around 0.1, now the average is 1.4
We now also receive a lot of CPU alerts for our environment that we just ignore (because they'll automatically be "fixed" within a few minutes).
Seeing that you have the exact same symptoms, I think something was changed in Mendix 5.17/5.18/5.19 that caused this.