failed to establish connection to the newly runner about kungfu HOT 5 CLOSED

instead of adding retry time for Control msg, I added a Wait API (#295), and called it before sending the Control msg.

I think this has two benefits:

1). we don't want to set retry time for every Control msg, some Control msg can fail by design.
2). Wait is a generic API that has a clear semantics, it can be used in conjunction with the Control msg or other APIs to achieve desired effect.

from kungfu.

@lgarithm , thanks for the reply, totally agree;

from kungfu.

hope this problem can be solved soon ~ as there is a high risk to trigger it when we scale up the job in kubernetes cluster, due to we can't keep the strict time sequence about making the kung-fu runner running and then update the config.json

from kungfu.

@zrss can you try the new master.
Default wait timeout is 5m, and can be set via
export KUNGFU_CONFIG_WAIT_RUNNER_TIMEOUT=10m

from kungfu.

@lgarithm , thanks, now it works well

from kungfu.