Allow the use of enable_failover
philwinder opened this issue · comments
AFAIK, enable_failover: true doesn't work. The framework should failover correctly.
I've just tested it. From what I can see it actually works.
Can you remind me about the intended behaviour? Maybe I misunderstand the command.
With failover enabled the framework will not deregister on Scheduler SIGTERM.
With failover disabled the framework will be deregistered on Scheduler SIGTERM.
What were your expectations?
Ah ok. The problem I was having was when logstash wasn't setup correctly. Given that you set enable_failover: true. When an executor failed for some reason, it wasn't restarted. Eventually there would be no executors at all and the scheduler wouldn't start any. What's worse is that when I restarted the scheduler, it still didn't restart any executors, like it had some bad state.
The way I got it to work was to set enable_failover: false, correct my mistakes in configuration, and then everything has been working ever since.