hyperledger fabric arquitecture issues
I've been recently assign the task to manage a Hyperledger Fabric network and investigate several issues related to timeouts. The problems occur somewhat randomly rendering the related application to simply fail. The only workaround at the moment has been to "delay" the transactions. Seems the network is not able to handle concurrency requests at all at times. Sometimes the network just stop running and the only fix to restore from backups!
The fatal error is:
connectToSomeEndpoint -> Could not connect to any endpoint of ... (list of orderers)
As a result I'm doing a full review of the network architecture and found the following design:
It seems wrong to me to have Orderer and Peer node under same organizations. And in addition to that, i'm not sure how the orderers interact among themselves by being each one on a different org.
Should I just plan to change the entire architecture?
It seems unlikely those are all CA servers. The (IBM) recommendation is a CA Server per node. (1 for each Peer and 1 for each Orderer). The names make no sense as CA servers.
Is it Kafka or Raft? In any event,only 1 Orderer is the leader. The others are for failover.
Those org names are from the testnet. Not really designed for robustness. Really only designed for a developer to use to write the chaincode.
I have seen similar when a network is running on VMs and not used for some time period. The VMs may get shut down and on restart they don't correctly restart all containers.