Greg Mann created MESOS-5635: -------------------------------- Summary: Agent failure during recovery prevents reregistration Key: MESOS-5635 URL: https://issues.apache.org/jira/browse/MESOS-5635 Project: Mesos Issue Type: Bug Reporter: Greg Mann
This issue was observed recently on an internal test cluster. Due to a bug in the agent code (MESOS-5629), regular segfaults were occurring on an agent. While the agent was recovering from one of these failures, it segfaulted again. After this time, we noticed that after recovery, the agent did not print {{Finished recovery}}, and its logs did not show any indication of reregistering with the master. Looking at the master's logs, however, the following line was observed repeatedly, at intervals on the order of seconds: {code} W0617 21:27:07.010679 2016 master.cpp:4773] Agent 2b899dd3-3b1f-4520-a6b2-98e32196f723-S4 at slave(1)@10.10.0.87:5051 (10.10.0.87) attempted to re-register after removal; shutting it down {code} These re-registration attempts had no corresponding lines in the agent log. Subsequently deleting the contents of the agent's {{work_dir}} and restarting it led to a successful registration with a new agent ID: {code} I0617 21:29:01.246119 2011 master.cpp:4635] Registering agent at slave(1)@10.10.0.87:5051 (10.10.0.87) with id 2b899dd3-3b1f-4520-a6b2-98e32196f723-S5 {code} -- This message was sent by Atlassian JIRA (v6.3.4#6332)