gc-log is within STW event?

[Thomas Schatzl]

Hi Ron,

On Mon, 2019-03-04 at 13:57 -0800, Ron Reynolds wrote:
> this blog 
> https://engineering.linkedin.com/blog/2016/02/eliminating-large-jvm-gc-pauses-caused-by-background-io-traffic
>  raises the question i my mind - why would logging to gc-log be done
> during a STW event at all?  wouldn't it make sense to defer gc-
> logging to after the STW event?  i suppose writing/flushing to gc-log 
> would allow one to monitor GC progress DURING a STW event but is that
> worth the risk of being suspended during a STW by IO operations?
> sorry if this is the wrong channel to ask such questions.
> .....................ron.
> _______________________________________________


  the GC log has traditionally been a tool for developer debugging,
also ease of adding new events (i.e. a "printf" is easy to add) for the
dev has been driving this. So can be quite important to not loose
information.

JFR has only been introduced much later for end-user use which
implements the writing out in a separate thread as suggested as far as
I know.

So JFR events are not as detailed as the GC log at this time for most
if not all collectors, but getting better with more recent releases.

My personal opinion would be to add more JFR events instead of fixing
the logging, but I think contributions in either direction are welcome.

Thanks,
  Thomas