So I am not sure why this is happening But ignition is maxing out the Ram that is has. We have tried several things including increasing RAM on the VM, Increasing processors, making the init ram equal to the max ram on the .conf file and doing this change
wrapper.java.additional.2=-XX:MaxGCPauseMillis=100.
None of it has seemed to much of anything.
What happens on a server restart is the ram and cpu will start at a reasonable number then it continues to build and drop but not by enough then it loops this until the RAM is maxed and we get the clock drift error. I thought it was the garbage collector not working correctly but maybe it is a memory leak and in which case I have no clue how to find it.
Thanks for the help.
This would be my bet, and they aren't easy to hunt down at all. Generally look for things that are creating background threads or utilizing persistence to try and maintain values across gateway start ups.
That being said, how much ram is available on the Gateway, and is there any other software besides Ignition running on this machine?
The gateway was just upped from 8-12 and there is 24 on the VM. As for what else is on the machine it isn't anything else really, just normal windows operating stuff ignition is using 13.5Gb and the next closest is a antimalware that is 250MB.
Unless you are highly familiar with java debugging tools, you probably want IA support to help you nail down where you are using/losing memory.
I am not I guess I will give them a call. Thanks for the quick responses guys.
You might check your logs and see if you're getting any other errors also. I had a similar issue a week ago and realized my database was having issues. (I don't recommend using RocksDB extension for MariaDB - it was causing issues with locking tables, so I switched to Timescale on Postgres).
I am getting other errors I guess I assumed they coincided with the RAM being maxed out. They are ABControlLogixTransportPool, ABControlLogixReadRequest and TimeoutDaemon. These all happen the same time the ClockDriftDetector error happens.
It is normal for a GC stall that is long enough to delay network responses for those drivers. Those are caused by the stall, not the cause of the stall.
I took a lunch and came back and the memory trend looks good. I didnt do anything except I realized the "wrapper.java.additional.2=-XX:MaxGCPauseMillis=100." was commented out of the .conf file. It is my understanding that memory leaks do not resolve themselves. After that change and restart it was still acting up but after 45min it seems to have resolved itself. Is this normal for a change to take 45 min to take effect or is this likely to be a problem again in the near future?
Without examining a heap dump captured during the problem, there's no good way to tell.
Ok nothing I can do now then thanks again for the help.