12-27-2016 01:02 PM
Here Again asking for some help about ERDAS Apollo.
I´m publishing a very large vector dataset from a postgresql /postgis connection. After setting up the layer through data manager apollo process started increasing CPU usage to the point that the entire server becomes unresponsive and apollo itself stop answering requests. I guess apollo is doing some configuration for the layer, but there is no info available about it. Is it normal? how long should that take in a normal scenario?
01-09-2017 12:24 PM
It’s a vector layer from legacy services and don’t think is a memory problem since it stays under 60%, CPU is what goes to 98% on a 12 core server and there it stays that way for about an hour. During and after the spike in CPU usage, apollo remains unresponsive and service needs to be restarted to get apollo back. If I restart the service during the spike, Apollo will be responsive a couple minutes while CPU usage gradually increases and then the server is stuck again. This happen only with very large datasets and remains me of how other servers will do similar things during cache generation, but that should not be the case,right?.
01-12-2017 06:39 AM
The APOLLO 2016 server actually starts 3 java processes – one for the main server (tomcat7.exe) then two child java.exe processes. The child processes are the RDS processes. The RDS processes are what does a lot of the imagery decoding/rendering, it is farmed out to these child processes for stability reasons and because one of them is 32-bit, for hosting the 32-bit Imagine libraries (the main server process and the other RDS process are both 64-bit). If the APOLLO server has started up correctly, you should see something like this in process explorer:
You can configure Tomcat memory: (Double click C:\Program Files\Hexagon\ERDAS APOLLO\tomcat\bin\APOLLOTOMCAT7w.exe)
You can configure RDS memory (for both 32bit and 64bit): While 32bit Java can only use 2GB memory: