[Building Sakai] has anyone seen one of these before? hibernate cache death.

[Jim Eng]

Thanks, Steve. I appreciate your response.

Yes, I saw jars with "hibernate" in their names, but these seem to be sakai code that uses hibernate in some way, and did not seem to be the pattern Ian was referring to.  

It turned out that for my immediate problem it's likely that something else caused a sakai shutdown for which the earliest warning or error messages was a set of messages for messageforums with the same pattern discussed in that thread from five years ago.  But the underlying problem was something else, and I do not think it will repeat. 

Unfortunately, after spending a little time trying to understand those symptoms, I do not have anything new to add to this thread that might be useful to someone who sees the same pattern in the future.

Jim  


On Dec 10, 2012, at 5:38 PM, Steve Swinsburg wrote:

> If I'm reading the old thread correctly, Ian was talking about a webapp war, not a component war:
> http://old.nabble.com/has-anyone-seen-one-of-these-before--hibernate-cache-death.-td9054857.html
> 
> Grepping for hibernate in the deployed webapp finds several references to hibernate, there are no deployed jars though:
> 
> [steveimac:~/dev/sakai/tomcat/trunk/webapps]$ find . -name hibernate\*
> [steveimac:~/dev/sakai/tomcat/trunk/webapps]$
> 
> cheers,
> Steve
> 
> 
> 
> 
> On 11/12/2012, at 3:25 AM, Jim Eng <jimeng at umich.edu> wrote:
> 
>> I have found only one hibernate jar among the artifacts deployed for this sakai instance (basically a trunk build from a month or so ago with some additional contrib projects and ctools-specific projects).  That is here:
>> 
>> 	shared/lib/hibernate-3.2.7.ga.jar
>> 
>> The other interpretation I thought of for Ian's message is that in some project some code is using a static hibernate method to get a session factory instead of getting it from the component manager.  I may have done that myself at some time.  Is that the meaning?  
>> 
>> Appreciate any advice anybody can give.  
>> 
>> Thanks.
>> 
>> Jim
>> 
>> 
>> On Dec 10, 2012, at 10:21 AM, Jim Eng wrote:
>> 
>>> 
>>> Ian:  This is reaching back to ancient times for you, I'm sure, but you gave
>>> the following advice five years ago.  I'm not sure I understand correctly,
>>> but it seems like the second paragraph may be saying that a hibernate jar is
>>> being deployed in some sakai component war instead of having that component
>>> rely on a common hibernate jar.  If that's the case, would it make sense to
>>> try to find the culprit by looking for a hibernate jar in a component war
>>> and then revising the poms for that project to avoid including the hibernate
>>> jar in the war? Would it make sense to think the cause of this must be a
>>> contrib or local project rather than part of sakai's kernel or core because
>>> otherwise it would be a common problem?
>>> 
>>> 
>>> 
>>> Ian Boston-2 wrote:
>>>> 
>>>> This is caused the ehcache singleton being bound to in a war and that 
>>>> war being reloaded.
>>>> 
>>>> Its often caused by having a Hibernate Session factory inside the war 
>>>> rather than in components or a secondary use of ehcache.
>>>> 
>>>> The really nasty thing about this, is that you may not get the error 
>>>> message until some point after the suspect war has reloaded as it 
>>>> doesn't happen until one of the consumers of ehcache tries to access a 
>>>> stale copy of ehcache.
>>>> 
>>> 
>>> -- 
>>> View this message in context: http://old.nabble.com/has-anyone-seen-one-of-these-before--hibernate-cache-death.-tp9054857p34779895.html
>>> Sent from the Sakai - Development mailing list archive at Nabble.com.
>>> 
>>> _______________________________________________
>>> 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"
>