can't auto reconnect
Opened this issue · 4 comments
- the init method in the _checkpid does not pass the hosts params
- every time reconnect use the _configure method, it is useless, because self._hosts
is changed . If the master can't connect for temporary, but recovered soon, you will never connect to the origin master.
Have you verified this against a running cluster?
On your first issue - the whole point is to avoid sharing the same set of sockets if the process was forked. I believe this is how the default redis pool works as well.
We always use one master/slave pair.The default pool does not have the hosts param, and always pass the connection_kwargs param over the init method when _checkpid.
Jondis will work only when the redis master has changed, but if the original master did not change(downtime is too short for failover or recover to the old master) , the _configure method won't work. Now we had to restart the web instance to resolve this problem.
OK, good to know. I'll take a look at this and see if I can reproduce the problem. I've been running Jondis in production for over half a year and haven't seen this yet though.