<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
<meta content="text/html; charset=ISO-8859-1"
http-equiv="Content-Type">
</head>
<body text="#000000" bgcolor="#ffffff">
<br>
<br>
On 03/01/11 09:26, Charles K Pepperdine wrote:
<blockquote
cite="mid:3BA76B61-E7DA-49E1-AEFE-819214B3C729@kodewerk.com"
type="cite">
<div>Tenured is 524288K of which 478087K is occupied, greater than
90% I've just been recently poking about in the source trying to
sort out how logs are being printed. I've seen that partial
fragments of the log are printed and that the " [CMS" fragment
is printed only after a CMS has been triggered. So, the problem
is bounded by the log messages (sans a stray pointer bug).</div>
<div><br>
</div>
<div>My question is, shouldn't a 90% occupancy of tenured trigger
a full GC?</div>
</blockquote>
<br>
The CMS concurrent collections are fast from what I've seen (on the
order of<br>
the time between ParNew collections). The rate at which objects
are getting<br>
promoted is also low (maybe 3m per ParNew collection). CMS thinks
it<br>
can wait to start a concurrent collection. The fact that a
"promotion failure"<br>
happened makes it look like fragmentation.<br>
<br>
<blockquote
cite="mid:3BA76B61-E7DA-49E1-AEFE-819214B3C729@kodewerk.com"
type="cite">
<div>
<div><br>
</div>
<div>Regards,</div>
<div>Kirk</div>
<div><br>
</div>
<div><br style="font-size: 15px;">
<div>
<div>On Mar 1, 2011, at 5:29 PM, Jon Masamitsu wrote:</div>
<br class="Apple-interchange-newline">
<blockquote type="cite">
<div>For 6692906 to be the problem there needs to be a<br>
CMS concurrent phase in progress (marking, precleaning
or<br>
sweeping) and a minor collection running (with<br>
UseParNewGC in use). From the fragment of the gc log<br>
I could not tell for sure (maybe it was in the ...
removed)<br>
but I don't think a concurrent phase was in progress<br>
so I would say it is not 6692906). Did you try<br>
-XX:-UseParNewGC as was suggested? Your minor<br>
pauses are not particularly long so maybe you<br>
could afford to try it. 6692906 will not happen<br>
without UseParNewGC. Note you need to turn of<br>
UseParNewGC as it is the default for CMS.<br>
<br>
Look back through the log for any other<br>
ParNew (promotion failed) and see what happens<br>
in those cases (in you find one). 2+ hours is too<br>
long.<br>
<br>
The gentleman who would know best about this code<br>
is out of the office until the end of the week. I'll
talk<br>
to him about this to see if he remembers a recent<br>
fix that I don't.<br>
<br>
<br>
On 03/01/11 01:38, Bogdan Dimitriu wrote:<br>
<blockquote type="cite">Hi guys,<br>
</blockquote>
<blockquote type="cite"><br>
</blockquote>
<blockquote type="cite">We're having a problem with
garbage collection as described here:<br>
</blockquote>
<blockquote type="cite"><a moz-do-not-send="true"
href="http://forums.oracle.com/forums/message.jspa?messageID=9345173">http://forums.oracle.com/forums/message.jspa?messageID=9345173</a>
(I<br>
</blockquote>
<blockquote type="cite">apologise if posting links is
not the right policy, but I prefer not to<br>
</blockquote>
<blockquote type="cite">duplicate data).<br>
</blockquote>
<blockquote type="cite"><br>
</blockquote>
<blockquote type="cite">We are going to try an upgrade
to JRE 6u24 soon, but reading the release<br>
</blockquote>
<blockquote type="cite">notes for each of the versions
since 6u20, I don't have much hope of<br>
</blockquote>
<blockquote type="cite">this upgrade fixing the problem.<br>
</blockquote>
<blockquote type="cite"><br>
</blockquote>
<blockquote type="cite">I have searched a bit on the
Java bugs database and I've come across<br>
</blockquote>
<blockquote type="cite">something that looks similar to
the problem I am experiencing:<br>
</blockquote>
<blockquote type="cite"><a moz-do-not-send="true"
href="http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=6692906">http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=6692906</a>,
but I'm not<br>
</blockquote>
<blockquote type="cite">convinced this is exactly the
same issue. This bug it seems will be<br>
</blockquote>
<blockquote type="cite">fixed in 6u25 (which I've read
will be released late March or early April).<br>
</blockquote>
<blockquote type="cite"><br>
</blockquote>
<blockquote type="cite">The reason I'm leaning towards
thinking this is a JVM bug is the fact<br>
</blockquote>
<blockquote type="cite">that the JVM can stay in the
hung state (as described on the forum) for<br>
</blockquote>
<blockquote type="cite">2+ hours until we kill the
process.<br>
</blockquote>
<blockquote type="cite"><br>
</blockquote>
<blockquote type="cite">I was hoping to get an idea
about this from the source :), so any hints<br>
</blockquote>
<blockquote type="cite">will be greatly appreciated.<br>
</blockquote>
<blockquote type="cite"><br>
</blockquote>
<blockquote type="cite">Thanks,<br>
</blockquote>
<blockquote type="cite">Bogdan<br>
</blockquote>
<blockquote type="cite">_______________________________________________<br>
</blockquote>
<blockquote type="cite">hotspot-gc-use mailing list<br>
</blockquote>
<blockquote type="cite"><a moz-do-not-send="true"
href="mailto:hotspot-gc-use@openjdk.java.net">hotspot-gc-use@openjdk.java.net</a><br>
</blockquote>
<blockquote type="cite"><a moz-do-not-send="true"
href="http://mail.openjdk.java.net/mailman/listinfo/hotspot-gc-use">http://mail.openjdk.java.net/mailman/listinfo/hotspot-gc-use</a><br>
</blockquote>
_______________________________________________<br>
hotspot-gc-use mailing list<br>
<a moz-do-not-send="true"
href="mailto:hotspot-gc-use@openjdk.java.net">hotspot-gc-use@openjdk.java.net</a><br>
<a moz-do-not-send="true"
href="http://mail.openjdk.java.net/mailman/listinfo/hotspot-gc-use">http://mail.openjdk.java.net/mailman/listinfo/hotspot-gc-use</a><br>
</div>
</blockquote>
</div>
<br>
</div>
</div>
</blockquote>
</body>
</html>