Production AEM 6.2 Author Servers spiked with high memory utilization

[Gunalan_V]

Our Production Author Servers are spiked with high memory utilization since three days.

I'm not seeing any errors in logs. But memory usage are something like below. Please let me know if some one have any suggestions!

Code Cache	61%	153.57MB	155.00MB	251.66MB	2.56MB	Non-heap memory
Metaspace	-17202010400%	172.03MB	193.81MB	-1.00B	0.00B	Non-heap memory
Compressed Class Space	1%	18.54MB	24.47MB	1.08GB	0.00B	Non-heap memory
PS Eden Space	10%	199.40MB	1.83GB	1.83GB	2.15GB	Heap memory
PS Survivor Space	0%	1.96MB	530.06MB	530.06MB	357.57MB	Heap memory
PS Old Gen	73%	4.23GB	5.73GB	5.73GB	5.73GB	Heap memory

GVK!

[Jörg_Hoh]

Can you post the headlines to that table as well?

Jörg

[Gunalan_V]

Hello Jorg,

Please find the header information as well.

Thanks,

GVK

[cquser1]

Going through the thread and head dumps will definitely provide you additional important details to track this down.

[Jörg_Hoh · Accepted solution]

Why do you think that this is a high memory utilization? You are looking at counters of the Java Virtual Machine, and without understanding this kind of memory model it is nearly impossible to judge if the numbers displayed there are good or not so good.

Do you have garbage collection enabled and can you see that you have full garbage collections? Do you have a consistent saw-tooth pattern? How much time is actually spent in garbage collection?

kind regards,

Jörg