Hi Reka, Thanks for the feedback. I've modified the code to print a warn log if the detected member id does not exist in the topology and remove it from the in-memory model. It will anyway remove the member id from in-memory model after publishing the member fault event. Fix commit [1].
[1] https://github.com/apache/stratos/commit/b80bfc4f6ae9a546b539676c3bc2a67fa56b106c Thanks. On Fri, Oct 30, 2015 at 11:15 AM, Reka Thirunavukkarasu <r...@wso2.com> wrote: > +1 for fixing it. CEP could print the WARN once and then it is better to > clean the in-memory data which contain the reference of the particular non > existent member in the Topology. > > Thanks, > Reka > > On Fri, Oct 30, 2015 at 12:27 AM, Akila Ravihansa Perera < > raviha...@wso2.com> wrote: > >> Hi, >> >> I'm seeing the following error message in Stratos after undeploying an >> application. >> >> Failed to publish member fault event. Member having [member-id] >> wso2brs-210-application.wso2brs-210-manager.wso2brs-210-manager.domainf0ff75ce-2f21-4ea9-9b5b-4299d761e64e >> does not exist in topology >> >> This is because when the application is undeployed, its members are >> removed from the topology but the health stats that were received >> previously are kept in CEP's memberTimestampMap. This map won't be cleaned >> until Stratos is restarted. >> >> I think if the member is not found in the topology when CEP detects a >> faulty member, it should just ignore that and remove the record >> from memberTimestampMap. I've created a JIRA at [1] to track this. >> >> >> [1] https://issues.apache.org/jira/browse/STRATOS-1594 >> >> Thanks. >> >> -- >> Akila Ravihansa Perera >> WSO2 Inc.; http://wso2.com/ >> >> Blog: http://ravihansa3000.blogspot.com >> > > > > -- > Reka Thirunavukkarasu > Senior Software Engineer, > WSO2, Inc.:http://wso2.com, > Mobile: +94776442007 > > > -- Akila Ravihansa Perera WSO2 Inc.; http://wso2.com/ Blog: http://ravihansa3000.blogspot.com
