Problem
Currently if Consensus client fails to synchronize or otherwise fatally fails, it can't notify the rest of the app, and the app keeps running. PR #772 fixes the most immediate problem. And RPC requests will still fail to be handled with outOfSync error which is good, but at given setup it can't force Helios to completely shut itself down or restart consensus client. So it just keeps running while being implicitly unusable.
Suggestions
We already have ConsensusSyncStatus enum and a channel exposing it but it is only used in wait_synced method. We could make Node check this status and make a force shutdown or just panic when it fails, potentially initiating several retries first.
Other suggestions welcome too.
Problem
Currently if Consensus client fails to synchronize or otherwise fatally fails, it can't notify the rest of the app, and the app keeps running. PR #772 fixes the most immediate problem. And RPC requests will still fail to be handled with
outOfSyncerror which is good, but at given setup it can't force Helios to completely shut itself down or restart consensus client. So it just keeps running while being implicitly unusable.Suggestions
We already have
ConsensusSyncStatusenum and a channel exposing it but it is only used inwait_syncedmethod. We could makeNodecheck this status and make a force shutdown or just panic when it fails, potentially initiating several retries first.Other suggestions welcome too.