----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/75325/#review227158 -----------------------------------------------------------
Ship it! Ship It! - Dineshkumar Yadav On Dec. 31, 2024, 10:17 a.m., sanket shelar wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/75325/ > ----------------------------------------------------------- > > (Updated Dec. 31, 2024, 10:17 a.m.) > > > Review request for ranger, Dineshkumar Yadav, Kishor Gollapalliwar, Abhay > Kulkarni, Madhan Neethiraj, Mehul Parikh, Pradeep Agrawal, Sailaja > Polavarapu, and Velmurugan Periasamy. > > > Bugs: RANGER-4520 > https://issues.apache.org/jira/browse/RANGER-4520 > > > Repository: ranger > > > Description > ------- > > Both usersync/tagsync instances becoming active and syncing users/tags in the > following scenario > Scenario: > Bring up 2 instances of UserSync so that one is active and other is passive > Cluster has 3 zk instances, stop 2 zk servers so that there is no quorum > Both usersync instances fail to establish Zk connection > Start both zk servers > Both usersync instances have transitioned to active and started performing > usersync > Same issue exists for tagsync also > Test: > Add user u1 to host1 > Add user u2 to host2 > Both users are synced and available on ranger admin side. > Usersync audits show details of both users synced > > EXPECTED BEHAVIOUR: > In case of usersync/tagsync running in HA mode, there should be only 1 > instance active at any point of time > CURRENT BEHAVIOUR: > After zk goes down and is back functional, more than 1 usersync/tagsync > instance becoming active > > > Diffs > ----- > > > ranger-common-ha/src/main/java/org/apache/ranger/ha/ActiveInstanceState.java > 6a7a0e7b8 > > > Diff: https://reviews.apache.org/r/75325/diff/2/ > > > Testing > ------- > > > Thanks, > > sanket shelar > >
