Changeable logic for response querying

[DieMyst]

• a client can customize the global timeout for the session
• exponential backoff on querying after some time
• add some config for sessions

Querying - periodic check that Tendermint's response is formed

[folex]

@DieMyst This task was created a while ago, could you please check that this issue is still relevant, and maybe actualize its description? I remember we had a few thoughts about websocket transport, wouldn't it supersede this task?

[DieMyst]

This is relevant and is not superseded by web sockets. We will support the HTTP protocol as well. And cannot use only push-from-server model in web sockets. Querying result process should remain because of these reasons.

This issue is about:

• not to DDOS Fluence nodes if there is a lot of requests
• stop querying if there is no response for some time
• additionally: add queue for querying responses

[folex]

Seems like this issue could be split into several separate issues.

not to DDOS Fluence nodes if there is a lot of requests

1. An improvement in Fluence JS library - exponential backoff

stop querying if there is no response for some time

Seems like it would be implemented in #638

additionally: add queue for querying responses

This is, too, an improvement in Fluence JS library? Or should that queue be on the Node/Statemachine side?

And original points:

a client can customize the global timeout for the session

Is this about client giving up too early on transaction? Not sure what timeout is meant here

add some config for sessions

Also not sure which config is meant here. Can you clarify, please?

[folex]

• Timeout & exponential backoff Implement timeout on tx send/query in Fluence JS #678
• Timeout from a 'server side' Mark transaction as failed if it's not processed after N blocks #638
• Batch response querying Batch query for tx responses #680
• Queue/bounded parallelism Limit tx response query parallelism #679

Closing this as superseded by the above tasks.