You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@syncope.apache.org by Anubhav Sharma <as...@talend.com> on 2015/04/22 19:03:02 UTC
Syncope cluster (multiple instance) deployment
Hi Guys,
We have the following deployment for Syncope:
Tomcat1 – Syncope1 – SinglePostgresDB
tomcat2 – Syncope2 - SinglePostgresDB
Both these tomcats are not in cluster. Syncope1 and Syncope1 are independent instances and not in cluster as well. Both these Syncope instances use a common single PostgresDB. Both the tomcats are behind a LoadBalancer and the application requests ELB for accessing Syncope so that load is randomly distributed.
With this deployment sometimes our application works!! But sometimes when the application tries to delete a user or change password, one of the Syncope throws an error:
09:58:21.999 DEBUG org.apache.syncope.core.rest.controller.ControllerHandler - Unresolved bean reference ...
09:58:22.156 DEBUG org.apache.syncope.core.rest.controller.ControllerHandler - After throwing UserController.search
09:58:22.158 ERROR org.apache.syncope.core.rest.utils.RestServiceExceptionMapper - Exception thrown by REST method: null
java.lang.NullPointerException
09:59:21.564 DEBUG org.apache.syncope.core.rest.controller.ControllerHandler - Before UserController.search([org.apache.syncope.core.persistence.dao.search.SearchCond@35f75285[
type=LEAF
subjectCond=<null>
attributeCond=org.apache.syncope.core.persistence.dao.search.AttributeCond@4ab571[
type=EQ
schema=accountName
expression=testAccount
]
And additionally we have warnings in the logs:
"message": "410373728 syncopePersistenceUnit WARN [http-bio-8080-exec-2529] openjpa.Runtime - Detected possible orphaned reference to deleted object \"[9899]\" in \"org.apache.syncope.core.persistence.beans.user.UAttr.values<element:class org.apache.syncope.core.persistence.beans.user.UAttrValue>\" of instance \"10764\".”,
After these errors, this Syncope instance always results error and if I try to login to Syncope-Console UI, I cannot connect to the Users tab….it results into error “Error in core”
Question is, is it a valid deployment scenario?? Is Syncope expected to work in such a deployment? Or we need to setup tomcat/Syncope in cluster and only then it is expected to work??
Can someone please provide hints!!!
Thanks,
Anubhav
Re: Syncope cluster (multiple instance) deployment
Posted by Francesco Chicchiriccò <il...@apache.org>.
On 22/04/2015 19:03, Anubhav Sharma wrote:
> Hi Guys,
>
> We have the following deployment for Syncope:
>
> Tomcat1 – Syncope1 – SinglePostgresDB
> tomcat2 – Syncope2 - SinglePostgresDB
>
> Both these tomcats are not in cluster. Syncope1 and Syncope1 are independent instances and not in cluster as well. Both these Syncope instances use a common single PostgresDB. Both the tomcats are behind a LoadBalancer and the application requests ELB for accessing Syncope so that load is randomly distributed.
>
> With this deployment sometimes our application works!! But sometimes when the application tries to delete a user or change password, one of the Syncope throws an error:
>
> 09:58:21.999 DEBUG org.apache.syncope.core.rest.controller.ControllerHandler - Unresolved bean reference ...
> 09:58:22.156 DEBUG org.apache.syncope.core.rest.controller.ControllerHandler - After throwing UserController.search
> 09:58:22.158 ERROR org.apache.syncope.core.rest.utils.RestServiceExceptionMapper - Exception thrown by REST method: null
> java.lang.NullPointerException
> 09:59:21.564 DEBUG org.apache.syncope.core.rest.controller.ControllerHandler - Before UserController.search([org.apache.syncope.core.persistence.dao.search.SearchCond@35f75285[
> type=LEAF
> subjectCond=<null>
> attributeCond=org.apache.syncope.core.persistence.dao.search.AttributeCond@4ab571[
> type=EQ
> schema=accountName
> expression=testAccount
> ]
>
> And additionally we have warnings in the logs:
>
> "message": "410373728 syncopePersistenceUnit WARN [http-bio-8080-exec-2529] openjpa.Runtime - Detected possible orphaned reference to deleted object \"[9899]\" in \"org.apache.syncope.core.persistence.beans.user.UAttr.values<element:class org.apache.syncope.core.persistence.beans.user.UAttrValue>\" of instance \"10764\".”,
>
>
> After these errors, this Syncope instance always results error and if I try to login to Syncope-Console UI, I cannot connect to the Users tab….it results into error “Error in core”
>
> Question is, is it a valid deployment scenario?? Is Syncope expected to work in such a deployment? Or we need to setup tomcat/Syncope in cluster and only then it is expected to work??
>
> Can someone please provide hints!!!
Hi,
you are missing an important piece of configuration, e.g. to let the two
OpenJPA instances (embedded in the two Syncope instances) synchronize
their respective data cache.
I have added some more information about this, including configuration
options, in the dedicated wiki page at
https://cwiki.apache.org/confluence/display/SYNCOPE/Setup+a+Syncope+cluster
HTH
Regards.
--
Francesco Chicchiriccò
Tirasa - Open Source Excellence
http://www.tirasa.net/
Involved at The Apache Software Foundation:
member, Syncope PMC chair, Cocoon PMC, Olingo PMC
http://people.apache.org/~ilgrosso/