<html>
<head>
<meta content="text/html; charset=UTF-8" http-equiv="Content-Type">
</head>
<body bgcolor="#FFFFFF" text="#000000">
Hi Michal (and John),<br>
<br>
I did look at Michal's log files (BTW, both are the same G1 logs).
And I can confirm that scan RS is the issue... here's the plot:<br>
<img src="cid:part1.05070800.01030408@oracle.com" alt=""
height="360" width="914"><br>
<br>
The plot above only shows the max obj copy times (for all pauses),
max RS Scan times and Parallel Time. So yes, scan RS is the culprit.
<br>
Also, looking at your logs, it seems that for mixedGCs the
reclaimable bytes don't cross 6%. So can you please try increasing
the G1HeapWastePercent to 10? (the latest builds will have 10 as the
default value).<br>
<br>
Please let me know if that improves your RT.<br>
<br>
-Monica<br>
<br>
<br>
On 1/31/2013 11:56 AM, John Cuthbertson wrote:
<blockquote cite="mid:510AB061.6010403@oracle.com" type="cite">Hi
Michal,
<br>
<br>
I haven't looked at the logs yet but from your description it
sounds like a large number of the RSets have been coarsened and/or
the fine grain tables have gotten very dense.
<br>
<br>
The RSet for a heap region tracks incoming references to objects
in that region. There are three levels on granularity: sparse,
fine, and coarse. The sparse and fine entries are encoded in open
hash tables based upon the region containing the references. As
the number of references from region A that point into region B
increase, the number of cards in the hash table entry for region A
increase (it's actually a bitmap with one bit per card) and as the
number of regions that contain references that point into B
increase, the number of region entries in fine grain table
increase. Once we run out of space in the fine grain table (i.e.
we can't add another bitmap for another region) we evict one of
the densest region bitmaps and say "coarsen" that entry. When we
have a coarsened entry we have to search the entire referencing
region looking for incoming references compared to searching
specific cards with the sparse and fine entries.
<br>
<br>
I'll take a look at your logs to see if I can confirm.
<br>
<br>
JohnC
<br>
<br>
<br>
On 1/31/2013 6:12 AM, Michal Frajt wrote:
<br>
<blockquote type="cite">Hi all,
<br>
<br>
After the iCMS got officially deprecated we decided to compare
the G1 collector with our best tuned (i)CMS setup. Unfortunately
we are not able to make the G1 young collection running any
closer to the ParNew. Actually we wanted to compare the G1
concurrent marking STW pauses with the CMS initial-mark and
remark STW pauses but already incredibly long running G1 young
collections are unacceptable for us.
<br>
<br>
We were able to recognize that the very long G1 young
collections are caused by the scanning remembered sets. There is
not much documentation about G1 internals but we were able to
understand that the size of the remembered sets is related to
the amount of mutating references from old regions (cards) to
young regions. Unfortunately all our applications mutate
permanently thousands references from old objects to young
objects.
<br>
<br>
We are testing with the latest OpenJDK7u extended by the 7189971
patch and CMSTriggerInterval implementation. The attached GC log
files represent two very equal applications processing very
similar data sets, one running the G1, second running the CMS
collector. The OpenJDK7u has an extra output of _pending_cards
(when G1TraceConcRefinement activated) which somehow relates to
the remembered sets size.
<br>
<br>
Young Comparison (both 128m, survivor ratio 5, max tenuring 15)
<br>
CMS - invoked every ~20 sec, avg. stop 60ms
<br>
G1 - invoked every ~16 sec, avg. stop 410ms !!!
<br>
<br>
It there anything what could help us to reduce the Scan RS time
or the G1 is simply not targeted for applications mutating
heavily old region objects?
<br>
<br>
CMS parameters
<br>
-Xmx8884m -Xms2048m -XX:NewSize=128m -XX:MaxNewSize=128m
-XX:PermSize=128m -XX:SurvivorRatio=5
-XX:MaxTenuringThreshold=15 -XX:CMSMarkStackSize=8M
-XX:CMSMarkStackSizeMax=32M -XX:+UseConcMarkSweepGC
-XX:+CMSClassUnloadingEnabled -XX:CMSWaitDuration=60000
-XX:+CMSScavengeBeforeRemark -XX:CMSTriggerInterval=600000
-XX:+UseParNewGC -XX:ParallelGCThreads=8
-XX:ParallelCMSThreads=2
<br>
<br>
G1 parameters (mind MaxNewSize not specified)
<br>
-Xmx8884m -Xms2048m -XX:NewSize=128m -XX:PermSize=128m
-XX:SurvivorRatio=5 -XX:MaxTenuringThreshold=15 -XX:+UseG1GC
-XX:MaxGCPauseMillis=200 -XX:G1MixedGCCountTarget=16
-XX:ParallelGCThreads=8 -XX:ConcGCThreads=2
<br>
<br>
G1 log file GC young pause
<br>
[GC pause (young) [G1Ergonomics (CSet Construction) start
choosing CSet, _pending_cards: 23697, predicted base time: 32.88
ms, remaining
<br>
time: 167.12 ms, target pause time: 200.00 ms]
<br>
[Parallel Time: 389.8 ms, GC Workers: 8]
<br>
>>>>
<br>
[Scan RS (ms): Min: 328.8, Avg: 330.4, Max: 332.6, Diff:
3.8, Sum: 2642.9]
<br>
<<<<
<br>
[Eden: 119.0M(119.0M)->0.0B(118.0M) Survivors:
9216.0K->10.0M Heap: 1801.6M(2048.0M)->1685.7M(2048.0M)]
<br>
<br>
Regards,
<br>
Michal
<br>
<br>
<br>
</blockquote>
<br>
</blockquote>
<br>
<div class="moz-signature">-- <br>
<a href="http://www.oracle.com" target="_blank"><img
src="cid:part2.08030404.02090907@oracle.com" alt="Oracle"
border="0" height="26" width="114"></a><br>
<font color="#666666" face="Verdana, Arial, Helvetica, sans-serif"
size="2">Monica Beckwith | Java Performance Engineer<br>
VOIP: <a href="tel:+1%20512%20401%201274"
moz-do-not-send="true">+1 512 401 1274</a>
<br>
Texas </font>
<br>
<a href="http://www.oracle.com/commitment" target="_blank"><img
src="cid:part5.00030004.08040308@oracle.com" alt="Green
Oracle" align="absmiddle" border="0" height="28" width="44"></a>
<font color="#4B7D42" face="Verdana, Arial, Helvetica, sans-serif"
size="1">Oracle is committed to developing practices and
products that help protect the environment</font>
<!-- This signature was generated by the MyDesktop Oracle Business Signature utility version 4.0 -->
</div>
</body>
</html>