Shenandoah and cometd testing

[Roman Kennke]

> On 4/9/19 11:44 PM, Simone Bordet wrote:
> > On Tue, Apr 9, 2019 at 11:13 PM Aleksey Shipilev <shade at redhat.com>
> > wrote:
> > FYI, in my Shenandoah run (1 server, 48g heap, 4 client machines at
> > each 1k clients at 10k messages/s) the max pause was about 4ms.
> 
> Cool. If you pass -Xlog:gc+stats, you might see the dissection of
> those pauses at the end of the
> run. For example, in a short run, -Xlog:gc+stats:file=stats.log gives
> this on cometd-server side:
>   
> http://cr.openjdk.java.net/~shade/shenandoah/cometd-server-gc-stats.log
> 
> Seems to me that until we finish the pending refactoring and/or
> concurrent class unloading story,
> -XX:-ClassUnloading might be beneficial for this workload.
> 
> -Aleksey
> 

-XX:-ClassUnloadingWithConcurrentMark is default, so it is not actually
unloading any classes, except at full-gc. Or am I missing something?

I see the 'System Purge' happened 2x. Is that full-gc-as-concurrent-gc, 
which also does class unloading? It looks like an accounting mistake.

Roman