> The output of "show controllers" is not always up to date for me and > lists already dead controllers forever. However, if I check the log, all > joining and leaving the group seems ok, only the show controllers > command outputs the died controllers. Seems like a simple not-in-synch > problem. > > Steps to reproduce: > > start controller A (192.168.10.90) > start controller B (192.168.10.92) > start controller C (192.168.10.94) > > kill controller B (killproc -g /usr/bin/java) > kill controller C > > `show controllers` on controller A outputs: > > --- > qualiproof-dev is hosted by 3 controller(s): > 192.168.10.94:1090 > 192.168.10.92:1090 > 192.168.10.90:1090 > > but should only contain 192.168.10.90:1090 anymore. > > I attached the log. See for more information. > > All 3 controllers find each other at startup. If I'm doing a graceful > shutdown or restart of a controller via /etc/init.d/ script, he updating > of the group-member list on "show controller" works correct. > > But if I do a forced simulated crash via `killproc -g /usr/bin/java` or > cut the network cable the output of "show controllers" will not always > be correct updated on all controllers. The dead controller remains > forever in the list of at least one remaining controller. If restart the > dead controller, the "dead" controller occurs even twice in the > group-member list of the other controller(s). > > > Sebastian >
Hello, the problem still exists for my configuration. Can someone reproduce this behavior? _______________________________________________ Sequoia mailing list [email protected] https://forge.continuent.org/mailman/listinfo/sequoia
