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

[Frajt, Michal]

Hi Damon,

It is not paging. The unpredictable delay from a scavenge to a CMS initial mark is the state of the current implementation. The CMSWaitDuration does not work correctly. Please find the details in the "CMSWaitDuration unstable behavior" post on hotspot-gc-dev mailing.

We are facing same issue with extremely long initial mark pauses. Just proper waiting for a scavenge (customized OpenJDK 6) has reduced every single initial-mark pause from 1200ms to 20ms only. 

Regards,
Michal


-----Original Message-----
From: hotspot-gc-use-bounces at openjdk.java.net [mailto:hotspot-gc-use-bounces at openjdk.java.net] On Behalf Of Damon Hart-Davis
Sent: Mittwoch, 22. August 2012 11:31
To: Srinivas Ramakrishna; Bernd Eckenfels
Cc: hotspot-gc-use at openjdk.java.net
Subject: Re: Why abortable-preclean phase is not being aborted after YG occupancy exceeds 50%?

Hi,

Could it be paging when not all of the JVM heap is (able to be) in physical memory at the same time?

Rgds

Damon


On 22 Aug 2012, at 09:03, Srinivas Ramakrishna wrote:

> I am not sure why, with regular CMS, we should have this kind of upredictable delay from a scavenge to a CMS initial mark.
> It must be OS/scheduling and load etc. which we cannot control (although a delay of 177 s seems excessive and must either mean
> that the CMS wait duration was exceeded or something like that.

_______________________________________________
hotspot-gc-use mailing list
hotspot-gc-use at openjdk.java.net
http://mail.openjdk.java.net/mailman/listinfo/hotspot-gc-use