Why abortable-preclean phase is not being aborted after YG occupancy exceeds 50%?

[Srinivas Ramakrishna]

Hi Bernd --

I've unfortunately forgotten the full history of this exchange (at least my
mailer has, and my own cache is nowadays oversubscribed and prone to
evicting objects somewhat too rapidly), so I'll answer only the questions
in the email:-

On Tue, Aug 21, 2012 at 1:06 PM, Bernd Eckenfels <
bernd.eckenfels at googlemail.com> wrote:

> Y. S. Ramakrishna wrote:
> >> An alternative workaround that might also work
> >> for you would be -XX:CMSWaitDuration=X
> > That should have been: -XX:CMSMaxAbortablePrecleanTime=X
>
> I have a question on this, can the –XX:+CMSScavengeBeforeRemark be
> combined with -XX:CMSMaxAbortablePrecleanTime=ms?
>

Yes, they can be combined.


>
> Because in my scenario the Scavenger Intervall is rather large
> (100-200s), so I dont want to use 400.000ms in the
> MaxAbortablePrecleanTime.
>

OK.


>
> The same is true for the initial-mark: when I use CMSWaitDuration, can
> I specify if should trigger a scavenger when timeout is reached? Or
> would it be OK to specify 400s?
>

Initial mark is typically scheduled immediately after a scavenge, so no
timeout
specification should be necessary. Perhaps I misunderstood yr question and
may be
you can elaborate a bit more on what you want to achieve?


>
> Greetings
> Bernd
>
> PS: in my case I know I need to resize the generations to get more
> frequent scavenger runs, however it is hard to push that into
> production on that particular system.
>

Yes, I am somewhat painfully aware of the limitations of tuning around CMS'
various
shortcomings and am looking forward to G1 being a panacea for those
problems :-)

-- ramki
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.openjdk.java.net/pipermail/hotspot-gc-use/attachments/20120821/e2200ce5/attachment.html