Check if devices of a peer are already used in a cluster on deployment
Closed this issue · 2 comments
mirenz1 commented
Current architecture design allows only one EDGAR on one peer managing all devices on that peer.
One instance of EDGAR can be used in a cluster only once.
If one EDGAR manages multiple devices, each device can be configured in different clusters, but a cluster can only be deployed if all devices managed by the same EDGAR are not deployed in another cluster.
- Implement gRPC endpoint in PeerManager to retrieve PeerState via an Action.
- Integrate PeerState gRPC endpoint in LEA and CLEO for displaying Peer health.
- actions::assign_cluster() marks peers as Blocked.
- Introduce actions::unassign_cluster() to be called in ClusterManager::delete_cluster_deployment(), which also marks peer as Available.
- A cluster cannot be deployed, if one of the devices managed by one EDGAR is already deployed in another cluster
- Update documentation (potentially into PeerState chapter)
- Remove PeerMessagingBroker-arrow from CARL architecture diagram
Derived from #208
Use peer state model as discussed in #7
- Add info of what is still missing to complete story #7
voelkera commented
current status
- Implement gRPC endpoint in PeerManager to retrieve PeerState via an Action.
- Integrate PeerState gRPC endpoint in LEA and CLEO for displaying Peer health.
- actions::assign_cluster() marks peers as Blocked.
- Introduce actions::unassign_cluster() to be called in ClusterManager::delete_cluster_deployment(), which also marks peer as Available.
- A cluster cannot be deployed, if one of the devices managed by one EDGAR is already deployed in another cluster
- Update documentation (potentially into PeerState chapter)
- Remove PeerMessagingBroker-arrow from CARL architecture diagram