Safe validator initial start
kyle-u410 opened this issue Β· comments
π Feature
Less down time when starting a validator.
Motivation
For network launch in testnet, starting too many validators at the same time has caused network halts. These halts are avoidable if the validator is synced when the bond_validator
call is made.
Implementation options
- Running a node with
start --validator
in the case bond has not been called will run the node as client watching forbond_validator
. As soon as that transaction is seen the node will shift to validator mode. - Allow bond_public transaction to be sent from validator when it catches up to head. This is effectively a script that runs the validator as a client and then when the client is close to tip issues the
bond_validator
In our testing, we start validators all at once all the time and we don't see network halts. What are the details of these halts?