Grouper Users - Open Discussion List

[Chris Hyzer <>]

Hey,

I upgraded hibernate to the latest stable in Grouper 2.1.  (trunk, obviously 
not released yet).  All junit tests pass, though obviously this is on the 
experimental side...

Can you load test it and see if the problem went away?  :)

http://www.internet2.edu/grouper/release/2.1.0/

(note, ldappc(ng) didn't build, not sure if you need that)

Thanks,
Chris

Ps. Fyi, here are the jars upgraded.  Note, I think we shouldn't just upgrade 
all jars in the api/ws/ui in this release so we can contain this change, and 
we shouldn't fix it if it isn't broken.  :)  We still need to adjust the 
maven configs...  TomZ?  :)

hibernate.jar: from: 3.2.6.ga, to: 3.6.7.Final
commons-collections: from: 0.1.0 , to: 3.2.1
javassist: new: 3.12.0.GA
slf4j-api: new: 1.6.2
asm: from: 1.5.3, to: 3.3.1
asm-util: from: 1.5.3, to: 3.3.1
asm-attrs: removed
hibernate-jpa-2.0-api: new: 1.0.1.Final
slf4j-log4j12: new: 1.6.2
cglib: removed

in jdbc samples:
mysql-connector-java-bin: from: 5.1.5, to: 5.1.17


-----Original Message-----
From: Okke Harsta 
[mailto:]
 
Sent: Tuesday, May 17, 2011 12:05 AM
To: Chris Hyzer
Cc: 

Subject: Re: Grouper peformance

Sure. Grouper is an essential part of our SURFconext infra, so we would 
gladly help you guys. What is the status of the upcoming 2.0 release? How do 
you want to go about this?

Rgds,
Okke 

On May 16, 2011, at 11:19 PM, Chris Hyzer wrote:

> We don't use StatelessSession, though I think if we redid our caching we 
> could look into it.  Its weird, I just got this error this weekend in my 
> development environment.  I restarted tomcat and it worked fine after that. 
>  We haven't upgraded Hibernate in a long time, I wonder if it would help... 
> would you be able to do a load test in a new Grouper version to see if it 
> is fixed if we did that?  I don't know whats involved, but it would 
> probably not be trivial... not sure if this would be for 2.0... :)
> 
> Thanks
> Chris
> 
> -----Original Message-----
> From: Chris Hyzer 
> Sent: Thursday, May 12, 2011 9:18 AM
> To: 'Okke Harsta'
> Cc: 
> 
> Subject: RE: Grouper peformance
> 
> Just wanted to forward this one to the list, there was an 
> 
> org.hibernate.AssertionFailure: possible non-threadsafe access to the 
> session
> 
> in WS (though I don't know why it wouldn't happen anywhere in the API) 
> while under load.  Anyone else encounter such a problem?
> 
> Thanks,
> Chris
> 
> -----Original Message-----
> From: Okke Harsta 
> [mailto:]
>  
> Sent: Thursday, May 12, 2011 2:24 AM
> To: Chris Hyzer
> Cc: Tom Zeller
> Subject: Re: Grouper peformance
> 
> Unfortunately not, I have checked all the grouper* logiles, but it does not 
> get logged. But the stacktrace in the mail below is the answer from WS. We 
> have ran the tests again with the following property value upped:
> 
> hibernate.max_fetch_depth             = 2
> 
> and the error percentage went down (but under heavy load did not disappear 
> entirely). Do you use a stateless session in the 
> edu.internet2.middleware.grouper.internal.dao.hib3.Hib3MembershipDAO.FindAllByGroupOwnerAndMemberAndField?
> 
> Okke
> 
> On May 11, 2011, at 7:02 PM, Chris Hyzer wrote:
> 
>> Hey, do you have the WS logs for this too that include the stack?
>> 
>> Thanks,
>> Chris
>> 
>> -----Original Message-----
>> From: 
>> 
>>  
>> [mailto:]
>>  On Behalf Of Tom Zeller
>> Sent: Wednesday, May 11, 2011 12:48 PM
>> To: Chris Hyzer
>> Cc: Okke Harsta
>> Subject: Fwd: Grouper peformance
>> 
>> Haven't seen this, forwarding over to Chris.
>> 
>> ---------- Forwarded message ----------
>> From: Okke Harsta 
>> <>
>> Date: Wed, May 11, 2011 at 1:46 AM
>> Subject: Grouper peformance
>> To: 
>> 
>> 
>> 
>> Hi Tom,
>> Just a quick question. We are doing load tests on our infrastructure
>> and we encounter (0.01 % of the time) an error in Grouper:
>> Could not execute grouper webservice request: [url:
>> https://engine:/grouper-ws/servicesRest/v1_6_000/groups]
>> [error: ] [http code: 500] [response:
>> <WsGetMembersResults><resultMetadata><resultCode>PROBLEM_GETTING_MEMBERS</resultCode><resultMessage>There
>> were 0 successes and 1 failures of getting members for groups.
>> </resultMessage><success>F</success></resultMetadata><responseMetadata><resultWarnings></resultWarnings><millis>110</millis><serverVersion>1.6.3</serverVersion></responseMetadata><results><WsGetMembersResult><resultMetadata><resultCode>EXCEPTION</resultCode><resultMessage>,
>> group:
>> WsGroupLookup[groupName=nl:surfnet:diensten:performance_test_team_493_445,
>> groupFindResult=SUCCESS], org.hibernate.AssertionFailure: possible
>> non-threadsafe access to the session,
>> Problem in HibernateSession: HibernateSession: isNew: true, isReadonly:
>> true, grouperTransactionType: READONLY_NEW,
>> Exception in list: (class [Ljava.lang.Object;), ByHqlStatic, query:
>> &apos;select ms, m from MembershipEntry as ms, Member as m where
>> ms.ownerGroupId   = :owner            and  ms.memberUuid  =
>> :member           and ms.memberUuid = m.uuid  and ms.enabledDb =
>> &apos;T&apos;&apos;, cacheable: false, cacheRegion:
>> edu.internet2.middleware.grouper.internal.dao.hib3.Hib3MembershipDAO.FindAllByGroupOwnerAndMemberAndField,
>> tx type: null, tx type: nullBind var[0]: &apos;Param (class
>> java.lang.String):
>> &apos;owner&apos;-&gt;&apos;0ddcb7a7b1db4591b569935b99a6ceed&apos;Bind
>> var[1]: &apos;Param (class java.lang.String):
>> It is only under very heavy load and the root cause appears to be:
>> http://forum.springsource.org/archive/index.php/t-43437.html
>> Did you or any of your colleagues ever encountered this?
>> Thanks,
>> Okke
>