[ http://jira.jboss.com/jira/browse/JBESB-638?page=all ] Kevin Conner updated JBESB-638: ------------------------------- Description: Kurt, can you raise the following with jUDDI and have it fixed? Thanks. There is a race condition in the jUDDI codebase which causes transient registry failures. The symptom of the race condition is the following error message E_busy (10400) The request cannot be processed at the current time. The Registry is currently unavailable. The class containing the race condition is org.apache.juddi.registry.local.Registry Every call to Registry.getRegistry() results in the recreation and initialisation of a global variable (registry). As the assignment of the global variable is made *before* the initialisation method is called there exists a window of opportunity whereby a caller can obtain a reference to an uninitialised registry. The consequence of this is that calls to RegistryEngine.isAvailable() can return false. I'm not sure why the RegistryEngine is being recreated on every call but the current code is unsafe (even if ignoring multiprocessor memory models). There should not be a global reference to the registry if it is a requirement to recreate it on every invocation. If it is not a requirement then the initialisation should really be occurring within a static block. was: Kurt, can you raise the following with jUDDI and have it fixed? Thanks. There is a race condition in the jUDDI codebase which causes transient registry failures. The symptom of the race condition is the following error message E_busy (10400) The request cannot be processed at the current time. The Registry is currently unavailable. The class containing the race condition is org.apache.juddi.registry.local.Registry Every call to Registry.getRegistry() results in the recreation and initialisation of a global variable (registry). As the assignment of the global variable is made *before* the initialisation method is called there exists a window of opportunity whereby a caller can obtain a reference to an uninitialised registry. The consequence of this is that calls to RegistryEngine.isAvailable() can return false. I'm not sure why the RegistryEngine is being recreated on every call but the current code is unsafe (even if ignoring multiprocessor memory models). There should be not be a global reference to the registry if it is a requirement to recreate it on every invocation. If it is not a requirement then the initialisation should really be occurring within a static block. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://jira.jboss.com/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ http://jira.jboss.com/jira/browse/JBESB-638?page=all ] Kurt Stam closed JBESB-638. --------------------------- Applied fix to put init in static block https://issues.apache.org/jira/browse/JUDDI-103 -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://jira.jboss.com/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ http://jira.jboss.com/jira/browse/JBESB-638?page=all ] Mark Little closed JBESB-638. ----------------------------- Fix Version/s: 4.2 Milestone Release 3 (was: 4.2) Resolution: Done -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://jira.jboss.com/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira