How to deploy a Daml Driver for Corda ledger in distributed mode

Tutorials and Guides From DA Support
,
1

How to deploy a Daml Driver for Corda ledger in distributed mode

Following the instructions of README.pdf packaged into the artifact bundle, one can easily deploy an environment for evaluation purposes. The shipped configuration enables local deployment mode by default. Distributed deployment assumes minor tailoring of config files on each node. Bootstrap script checks a bunch of components to be ready (including docker-compose, postgresql, java: currently only up to Java version 1.8 accepted). The script generates a directory /tmp/deployment where network scripts will take place.

Demo deployment architecture

We have 3 different types of nodes: a Site node, a Notary node, and an NMS node.

Site node

  • A ledger API.
  • An associated IndexDB, that the ledger API needs to be able to talk to.
  • A Corda node.
  • An associated Vault DB, that the Corda subsystem needs to be able to talk to.

Notary node

  • A corda node specially configured as a notary. Every Corda node needs to be able to talk to this.
  • An associated Vault DB.

NMS node

  • An instance of Cordite . Every Corda node needs to be able to talk to this when it turns on.

There are multiple connections between nodes that need to be properly configured and checked.

DD4C - demo deploy
DD4C - demo deploy1720×1282 123 KB

If any of those lines of communication are broken, the Ledger API isn’t going to come up healthy. The configs exist to configure those lines of communication.

No ports are exposed from the containers

Distributed deployment assumes proper network connection among hosts involved in, otherwise the following erroneous Docker output might occur:

$ docker ps --format "table {{.Names}}\t{{.Status}}\t{{.Ports}}"
NAMES STATUS PORTS
nodenode3-tribeca-corda Up 3 minutes (health: starting)
nodenode3-tribeca-ledgerapi Up 3 minutes (health: starting)
nodenode3-tribeca-index-db Up 3 minutes (healthy)
nodenode3-tribeca-vault-db Up 3 minutes (healthy)

And the containers are going down after sometime with error

[INFO] 2021-03-24 03:09:19.798 - Retrying Corda Subsystem (Attempt 50) ...
[ERROR] 2021-03-24 03:09:29.800 - Corda subsystem unreachable before timing out, exiting.

Proper network setup of featuring hosts

Check logfiles that could be involved in:

/host3/deployment/Nodenode3-tribeca/ledgerapi/logs/20210325_125554.143550495_init.log
/host3/deployment/Nodenode3-tribeca/corda/logs/20210325_125553.806181382_init.log
/host2/deployment/Nodenode2-moorgate/ledgerapi/logs/20210325_125535.245255904_init.log
/host2/deployment/Nodenode2-moorgate/corda/logs/20210325_125534.960059960_init.log
/host1/deployment/NetMan/nms/logs/20210325_125515.927446489_init.log
/host1/deployment/Nodenode1-pembroke/corda/logs/20210325_125515.139796735_init.log

Look for ERROR messages in the logs as above:

Corda log

[ERROR] 2021-03-25 13:24:22.961 - NMS Service unreachable before timing out, exiting.

Ledger API log

[ERROR] 2021-03-25 13:03:55.576 - Corda subsystem unreachable before timing out, exiting.

To test network connectivity, you could ssh into the host running node2 or node3 and run

​curl < *ip_NMS* >:< *port_NMS* >

to reach node0 NMS.

Possible causes

  • Firewall settings on either host are preventing communication.
  • There’s an issue with the configuration.
  • There’s a bug with the bootstrapper.
  • Its databases didn’t come online.
  • The Notary (Pembroke) was not reachable.
  • The NMS service was not reachable.
4 Likes