restart strategy

[InoMurko]

Overview

Increasing the restart strategy.

Changes

Testing

/

[boolafish]

wow...copy a explanation from slack by Ino:

---

our etheruem connection monitor checks the connection every ~8 seconds. so the connection monitor didn’t catch the network glitch and it didn’t raise a ethereum_connection_error alarm.
the processes that are dependent to a working ethereum connection started crashing, but the crash limit on the supervisor was really low (3 crashes in the 5 second time frame). so tthat means the supervisor will crash itself:

    # Assuming the values max_restarts and max_seconds,
    # then, if more than max_restarts restarts occur within max_seconds seconds,
    # the supervisor terminates all child processes and then itself.
    # The termination reason for the supervisor itself in that case will be shutdown.
    # max_restarts defaults to 3 and max_seconds defaults to 5.

[boolafish]

HeightMonitor should catch that in max 8 seconds and raise an alarm.

But you're still setting this to max 5 seconds. Is it okay to not be > 8?

[InoMurko]

the thing is -the timer resets itself if you don't reach max_restarts in max_seconds. so it's a bit difficult to figure out what the correct numbers are - for example, a delay in a process init/1 could reset the timer I'm mentioning.

[boolafish]

the timer resets itself if you don't reach max_restarts in max_seconds.

oh wow.....I see. I thought if force failure as long as it runs overtime.