Need a feature to deal with network isolation between capture and tikv node

[maxshuang]

Is your feature request related to a problem?

Currrently implementation show that ticdc region always find region leader to subscribe real-time data.
When network isolation occur between a specified capture and tikv node(region leader), but pd cluster and tikv cluster are all good, this capture will fail to sync the real time data and fail to do some failover work, like moving the job to other capture.

It's unlikely to happen now because the tikv cluster and cdc cluster are deployed in the same DC. But when we are in a cloud environment, things maybe totally different.

So we record the new feature to deal with this abnormal scene.

Describe the feature you'd like

Automatic job failover when network isolation occur

• Processor can sense the network isolation between a specified capture and tikv node(region leader), then report to pd
• Owner can move the specified fail job to other capture

Describe alternatives you've considered

No response

Teachability, Documentation, Adoption, Migration Strategy

No response