Hung JVM consuming 100% CPU
Y. Srinivas Ramakrishna
y.s.ramakrishna at oracle.com
Thu Mar 3 09:57:11 PST 2011
On 3/3/2011 4:31 AM, Bogdan Dimitriu wrote:
> Thanks Srinivas,
> We will upgrade our servers to JRE 6u24 soon and I will let you know after a while if we are still
> seeing the problem.
> If it is still a problem, do you reckon that fiddling with the -XX:CMSInitiatingOccupancyFraction
> option could improve the fragmentation issue and hence will make compaction less likely (assuming
> the bug is in the compaction code)? We've got some CPU cycles to spare, so we could afford to lower
> this value a bit.
The bug i am thinking about is actually not in compaction code; rather it's in ParNew+CMS scavenge
code that causes promoted objects to be corrupted. Compaction is then unlucky to run into those
objects (because they are still alive). So avoiding the compaction is not necessarily going to
avoid the issue.
Nonetheless, avoiding a situation where compaction is needed is probably worthwhile, because
I am assuming that you are using CMS so as to avoid the kind of costly GC pauses that compaction
entails. Thus, tuning to avoid compaction has independent value, i think.
PS: By the way, I think it might be worthwhile to send in the core file to Oracle, in any case,
for analysis to see if the crash in question has the rough shape of the bug(s) that we
were thinking about...
> On 03/03/11 11:01, Y. Srinivas Ramakrishna wrote:
>> Hi Bogdan --
>> I agree with Jon that this is not 6692906.
>> Rather it's a case of a bug in object size computation
>> causing the GC to go awry. There _were_ a few related
>> bugs fixed in this area not long ago (in 6u23 i believe),
>> so I believe you'll find that the upgrade to 6u23
>> or higher would fix this problem.
>> If you find that the latest 6uXX update (6u24?) still
>> exhibits this problem, please file a new bug via your
>> Java support channel and let us know.
>> -- ramki
>> On 3/1/2011 1:38 AM, Bogdan Dimitriu wrote:
>>> Hi guys,
>>> We're having a problem with garbage collection as described here:
>>> http://forums.oracle.com/forums/message.jspa?messageID=9345173 (I
>>> apologise if posting links is not the right policy, but I prefer not to
>>> duplicate data).
>>> We are going to try an upgrade to JRE 6u24 soon, but reading the release
>>> notes for each of the versions since 6u20, I don't have much hope of
>>> this upgrade fixing the problem.
>>> I have searched a bit on the Java bugs database and I've come across
>>> something that looks similar to the problem I am experiencing:
>>> http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=6692906, but I'm not
>>> convinced this is exactly the same issue. This bug it seems will be
>>> fixed in 6u25 (which I've read will be released late March or early April).
>>> The reason I'm leaning towards thinking this is a JVM bug is the fact
>>> that the JVM can stay in the hung state (as described on the forum) for
>>> 2+ hours until we kill the process.
>>> I was hoping to get an idea about this from the source :), so any hints
>>> will be greatly appreciated.
>>> hotspot-gc-use mailing list
>>> hotspot-gc-use at openjdk.java.net
hotspot-gc-use mailing list
hotspot-gc-use at openjdk.java.net
More information about the hotspot-gc-dev