We experienced a higher level of load and the dynatrace server started to enqueue in the RTA queue till it got full and started to drop purepaths. We are already on XLarge96 and our machine is a 40CPU with 160GB RAM. During the incident the CPU level was never more than 40% and the memory was far to be full.
Is there any way to tune the server to better use our machine? The RTA threads are currently 10, is there any way to force the RTA to use more?
Solved! Go to Solution.
Hi, yes you can force it to use more entering debug mode > server settings > general > debug options, and giving server flag com.dynatrace.diagnostics.server.realtime.maxWorkerThreads where value is number of threads. I have the same sizng at xLarge96 and am at 24 rta threads
I have encountered similar and would advise you to checkout the following if this is going to be an ongoing problem: any expensive RegEx's - find the output of time and values in the server.log. Also look at if you can further optimise data and measures e.g. those frequent exceptions, aggregation in JDBC and ADO if needed, and wildcard captures in sensor properties can be expensive.
Hope it helps
Hi Gary, thanks for your answer. I'm now working on the RegEx tuning but I've found that in the logs, the RegExExecutionTimeStatistics report is for the last 4 weeks.
2017-10-08 23:09:50 INFO [RegExExecutionTimeStatistics] Summary of most significant Regular Expressions since 09/13/17 11:09:50 PM
Is there any way to have it more often? Any other tool to tune the RegEx?
Hi Cristian, yes by default it does not run so regularly. You can set debug flag com.dynatrace.diagnostics.regexperformance.debug.interval where the value is in minutes. We have ours set to 360 so get some data every 6 hours. In terms of tuning RegEx there are a few sites out there like regex101 you can test with and some advice if you search around the community too - I will always try to avoid forward/backward lookups in long strings and wildcard etc maybe that helps you too!