JVM Out of Memory - Cluster manager failed

Hi Mendix Community, We get the following error message in the log when deploying our application. After login the system hangs and then comes back with an error. Jul 15 13:54:31.020 tr10000: INFO - WebUI: Login OK: user 'MxAdmin' (Number of concurrent sessions: 1). Jul 15 13:58:54.119 tr10000: WARNING - ConnectionBus_Queries: (1/4) Query executed in 16 seconds and 374 milliseconds: SELECT DISTINCT "system$clustermanager"."system$statisticsid", Jul 15 13:59:06.501 tr10000: WARNING - ConnectionBus_Queries: (2/4) "system$clustermanager"."system$xasinstanceid" Jul 15 13:59:09.344 tr10000: WARNING - ConnectionBus_Queries: (3/4) FROM "system$clustermanager" Jul 15 13:59:15.017 tr10000: WARNING - ConnectionBus_Queries: (4/4) WHERE "system$clustermanager"."system$statisticsid" = 1970324836974593 Jul 15 13:59:20.839 tr10000: JVM Out of Memory error detected Does anyone know what could be causing this? Thanks.
2 answers

In short: no. What you're seeing is that a random action (in this case the login) is triggering an out of memory error. These occur when the JVM (the java virtual machine, which hosts the Mendix runtime) tries to do 'something' and there's no available memory.

This 'something' does not necessarily take a lot of memory. It just happens to trigger it. In other words, imagine the JVM as a big bucket of water. Your login action was the drop that caused it to overflow, but it was probably not what filled it up.

Trying to figure out out of memory errors is a whole science in itself. That being said, they are often (but not always) caused by 'expensive' computations such as batch jobs, scheduled events or Microflows that operate on a large amount of data. You can investigate this by checking the running requests and try to detect a pattern. Alternatively, checking your graphs (trends) in the cloud portal if you're running there could help correlate the errors and events.


Seems like that on login of MxAdm a select is done on system$clustermanager, which has a huge impact on the memory, causing a heap space. Is there a login microflow or hopepage which is doing something special?