reconnect feature request

[virtser]

In case that I set a watch on some key.
What if etcd server is down or connection to server was lost?
How does the code reconnect?
In that case my application won't be notified of key value change.

[drusellers]

Agreed. I am going to define 'reconnect' for this issues as reconnecting to the same initially provided server. Etcetera needs to expose a timeout handler so that a user can 'rewatch' or take some other action as necessary.

[drusellers]

If my understanding of the TPL is correct - we can return a task on watch commands and then you could set handlers for the timeout / error / success - etc

[shamil]

Is this still planned?
Also if I work with multi-machine etcd cluster, how can I specify all of them when creating client?

[drusellers]

A) Yes
B) Currently, you can't - we have been thinking about how to best approach this as well. This is planned though.

[smalldave]

If it's of any use I started on this here
https://github.com/smalldave/etcetera/tree/noderetry
Stalled on the tests and unfortunately deadlines that pay the bills have stopped me going any further with it for a bit.