Lots of org.HdrHistogram.Histogram

[laurentlhermine]

Hi

I have a Spring Boot application calling several external endpoints through HystrixFeign (Feign 9.7.0 Hystrix 1.5.18 Spring 1.5.4) The app is suspected to have a memory leak. Heap dump analysis has led me to one suspect with objects of class org.HdrHistogram.Histogram which are instantiated by Hystrix. Their number is growing over time, after letting the memory drift for some time I have a heap dump with 10000 such objects that were not garbage collected taking 170Mb.

The config is as follows : hystrix: command.default.execution.isolation.thread.timeoutInMilliseconds: 60000 threadpool.default.coreSize: 20 threadpool.default.maxQueueSize: 20 threadpool.default.queueSizeRejectionThreshold: 20

I’d like to understand what drives instantiation of such objects and what will make them garbage collectable or not. Is it possible to estimate the maximum size they will take ? Is it possible that my code, configuration or a specific external condition leads to keeping these objects alive ?

Thanks

[Uil-Nella]

Also encountered a similar problem, can you @mattrjacobs @benjchristensen @dmgcodevil @quidryan @mebigfatguy help to have a look, at present is the adjustment of metricsRollingStatisticalWindowInMilliseconds configuration , temporarily solve this problem.

[msosto]

any updates on this issue?