support multi endpoint

[Revolution1]

• auto retry between endpoint
• auto balance between endpoint

[gszpura]

Hi,
some questions:

1. will there be an option to choose endpoint and switch only when endpoint failure occurs?
2. what's the timeline for this feature? When will it be ready?

[Revolution1]

@gszpura Of course, that is exactly what i'm going to do

As for the timeline, since I thought it's not a urgent feature, currently i'm working on a demo app of etcd3-py. If you truly need it, maybe I can write it first.

Also, thank you for using this lib and welcome to contribute to this project 😀

[gszpura]

Reconnection is needed for my usecase since my etcd HA cluster consists of 3 nodes and I would like to auto switch nodes when current one is failing.

If you have an idea and time for writing that I would appreciate it ;]

[stupidchen]

@gszpura Of course, that is exactly what i'm going to do

As for the timeline, since I thought it's not a urgent feature, currently i'm working on a demo app of etcd3-py. If you truly need it, maybe I can write it first.

Also, thank you for using this lib and welcome to contribute to this project 😀

@Revolution1 Are there any contribution guide about this project? I want to do something about this feature. Could you give me some advice?

[Revolution1]

@stupidchen you can just write your thoughts about this feature under this issue

I have some ideas about this feature:

Add a EnpointManager to control which endpoint should the client use.

Discover endpoint:

• user provide
• by dns resolve
• by etcd3 member_list api

Auto select a heathy endpoint

• switch after connection fails
• active health-check

Of course you can just start coding and create a pull request 😆

The only guide is write clean and readable code (and PEP8 formatted) .

ps: don't forget the test~

[stupidchen]

@stupidchen you can just write your thoughts about this feature under this issue

I have some ideas about this feature:

Add a EnpointManager to control which endpoint should the client use.

Discover endpoint:

• user provide
• by dns resolve
• by etcd3 member_list api

Auto select a heathy endpoint

• switch after connection fails
• positive health-check

Of course you can just start coding and create a pull request 😆

The only guide is write clean and readable code (and PEP8 formatted) .

ps: don't forget the test~

Thanks for your advice! I will start coding in few days!

[dariko]

I've submitted #55 which gives BaseClient the ability to failover between endpoints

[boylegu]

How does the current version support clustering? No solution？Release 0.2.0…. It seems so far off in the future