[Building Sakai] Java 1.6 for Sakai 2.7

[Charles Hedrick]

Actually I'm going to issue a service request with Sun, but our  
support contract was inadvertently changed to a level where I can't  
report problems. We're working on it with Sun.

There are lots of reports of GC issues with Java, some that look the  
same, but it's always a bit unclear what is causing them. If you look  
at release notes you'll see many patches involving the GC. It's quite  
likely that there are multiple issues resulting in similar symptoms.

This could well be an interaction between Java and Solaris, so you  
might not see it. One question is whether you look for it. I have a  
system status page that reports all kind of operational numbers. One  
of them is the last 3 GC's that took longer than 10 sec. (I get that  
by using a program that effectively tails catalina.out and watches  
matching reporting the amount of time applications were stopped, but  
can cope with the file changing when we do a new deploy.)


On Oct 9, 2009, at 10:23 AM, Noah Botimer wrote:

> Do we have any contacts with outside experts with expertise on  
> really big JVMs with a lot of churn? There have to be some other  
> apps that have run into this or some insight into what we could fix.
>
> Thanks,
> -Noah
>
> On Oct 9, 2009, at 9:52 AM, Charles Hedrick wrote:
>
>> I'd like to note that we're just about finished moving our systems  
>> from Java 6 back to Java 5. I'll update the list next week on the  
>> details, because until we've survive Monday I won't be certain that  
>> we've solved our problem. But the reason for the change is  
>> performance problems with the Java 6 GC that we think aren't  
>> present in Java 5. I'm very skeptical that these will be fixed, so  
>> I'm currently expecting that we'll need to stay with Java 5 until  
>> GC1 is ready for production (which it currently is not -- I've  
>> tried it).
>>
>> On Oct 8, 2009, at 5:26 PM, Seth Theriault wrote:
>>
>>> Seth Theriault wrote:
>>>
>>>> I just tried to build again with an updated trunk, but it failed
>>>> due to a repeated method in the Rwiki code. I'll be sending a
>>>> separate e-mail about that.
>>>
>>> Just finished up some local testing using trunk, r67389, on a Mac
>>> OS X Leopard system:
>>>
>>> $ uname -a
>>> Darwin mirth.cc.columbia.edu 9.8.0 Darwin Kernel Version 9.8.0:
>>> Wed Jul 15 16:55:01 PDT 2009; root:xnu-1228.15.4~1/RELEASE_I386 i386
>>> $ echo $JAVA_HOME
>>> /System/Library/Frameworks/JavaVM.framework/Versions/1.6/Home
>>> $ $JAVA_HOME/bin/java -version
>>> java version "1.6.0_15"
>>> Java(TM) SE Runtime Environment (build 1.6.0_15-b03-226)
>>> Java HotSpot(TM) 64-Bit Server VM (build 14.1-b02-92, mixed mode)
>>>
>>> Using the default pack-demo target, the build was without
>>> incident. Of course, the default uses "-Dmaven.test.skip=true" so
>>> I turned that off (and "-Dmaven.test.failure.ignore=true" too)
>>> and rebuilt.
>>>
>>> With the tests on, none of them fails but one of them returns
>>> JDBCerrors:
>>>
>>> [exec] [INFO]
>>> ------------------------------------------------------------------------
>>> [exec] [INFO] Building coursemanagement-hibernate-impl
>>> [exec] [INFO]    task-segment: [clean, install, sakai:deploy]
>>> [exec] [INFO]
>>> ------------------------------------------------------------------------
>>> ...
>>> [exec] Running
>>> org.sakaiproject.coursemanagement.test.CourseManagementAdministrationTest
>>> [exec] 16:48:58,732  WARN JDBCExceptionReporter:77 - SQL Error:
>>> -104, SQLState: 23000
>>>
>>> I'll note here that this test is imperfect because search is
>>> being downloaded as an "assembly" like kernel.
>>>
>>> With the sakai-demo package built, I was able to start it up.
>>> Initially, there are errors in blogger, chat, and samigo:
>>>
>>> $ grep ERROR logs/catalina.out
>>> 2009-10-08 16:31:02,974 ERROR main
>>> org.apache.catalina.core.ContainerBase.[Catalina].[localhost].[/ 
>>> sakai-blogger-tool]
>>> - Exception sending context initialized event to listener
>>> instance of class com.sun.faces.config.ConfigureListener
>>> 2009-10-08 16:31:02,977 ERROR main
>>> org.apache.catalina.core.StandardContext - Error listenerStart
>>> 2009-10-08 16:31:02,977 ERROR main
>>> org.apache.catalina.core.StandardContext - Context
>>> [/sakai-blogger-tool] startup failed due to previous errors
>>> 2009-10-08 16:31:04,518 ERROR main
>>> org.apache.catalina.core.ContainerBase.[Catalina].[localhost].[/ 
>>> sakai-chat-tool]
>>> - Exception sending context initialized event to listener
>>> instance of class com.sun.faces.config.ConfigureListener
>>> 2009-10-08 16:31:04,522 ERROR main
>>> org.apache.catalina.core.StandardContext - Error listenerStart
>>> 2009-10-08 16:31:04,522 ERROR main
>>> org.apache.catalina.core.StandardContext - Context
>>> [/sakai-chat-tool] startup failed due to previous errors
>>> 2009-10-08 16:31:37,888 ERROR main
>>> org.apache.catalina.core.ContainerBase.[Catalina].[localhost].[/ 
>>> samigo]
>>> - Exception sending context initialized event to listener
>>> instance of class com.sun.faces.config.ConfigureListener
>>> 2009-10-08 16:31:39,136 ERROR main
>>> org.apache.catalina.core.StandardContext - Error listenerStart
>>> 2009-10-08 16:31:39,136 ERROR main
>>> org.apache.catalina.core.StandardContext - Context [/samigo]
>>> startup failed due to previous errors
>>>
>>> but these clear up nicely if you add the
>>> "-Dsun.lang.ClassLoader.allowArraySyntax=true" to the JAVA_OPTS
>>> as Matt suggested.
>>>
>>> Seth
>>>
>>> _______________________________________________
>>> sakai-dev mailing list
>>> sakai-dev at collab.sakaiproject.org
>>> http://collab.sakaiproject.org/mailman/listinfo/sakai-dev
>>>
>>> TO UNSUBSCRIBE: send email to sakai-dev-unsubscribe at collab.sakaiproject.org 
>>>  with a subject of "unsubscribe"
>>
>> _______________________________________________
>> sakai-dev mailing list
>> sakai-dev at collab.sakaiproject.org
>> http://collab.sakaiproject.org/mailman/listinfo/sakai-dev
>>
>> TO UNSUBSCRIBE: send email to sakai-dev-unsubscribe at collab.sakaiproject.org 
>>  with a subject of "unsubscribe"
>

-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 2421 bytes
Desc: not available
Url : http://collab.sakaiproject.org/pipermail/sakai-dev/attachments/20091009/ed319899/attachment.bin