Alive and broadcast messages were interfering because prefixes were wrong. To get only the keys starting with "msg", we need to use a range end "msh" as 'h' is 'g'+1.
This is using etcd leases on "alive" keys to determine peer connectivity on a best-effort basis while ensuring and reporting overall network connectivity.
Signed-off-by: Sasha Bogicevic <[email protected]>
Signed-off-by: Sasha Bogicevic <[email protected]>
Signed-off-by: Sasha Bogicevic <[email protected]>
Signed-off-by: Sasha Bogicevic <[email protected]>
Signed-off-by: Sasha Bogicevic <[email protected]>
Signed-off-by: Sasha Bogicevic <[email protected]>
This is not properly handling errors and connectivity topics, but basic broadcast and delivery works.
These are not in use anymore and the Ouroboros/Reliability implementation is (currently) superseded by the Etcd-based implementation.
Backport of the HTTP chain observation to build a docker image of `hydra-chain-observer` for version `0.19` and deploy it to the explorer. This could be released as `0.19.1` --- * [ ] CHANGELOG updated or not needed * [ ] Documentation updated or not needed * [ ] Haddocks updated or not needed * [ ] No new TODOs introduced or explained herafter