Occasional startup failure when the network map node is started in parallel with others. about corda HOT 3 CLOSED

Here is a summary of the problem:

Symptoms:
Nodes will not startup except the node running the network map service.

How to reproduce the problem:
Setup: Node A, Node B, Notary(with Network Map Service)
There are 2 ways to reproduce the problem.

1. Startup the nodes in this sequence quickly (node A, node B, Notary), sometime Node A and Node B will fail to start.

2. Startup all the node normally (Notary -> NodeA -> Node B) and allow time between each startup.
   Nodes should start without any problem. Now shutdown all nodes and start them up again one by one. Node A and Node B will no longer be able to connect to the Notary.

Workaround:
Remove artemis folder and persistence.mv.db from the node hosting the network map service and restart the nodes will fix the problem.

from corda.

Found the bug!

In ArtemisMessagingServer we add a call back handler to the artemis server to create bridges for the queues. However the handler registration happen after the queues restoration, so all the existing queue won't have a bridge after server restart.

The fixes is in this branch, https://github.com/corda/corda/tree/pat-node-startup-fix

Will raise a PR after written a unit test for this.

from corda.

The fix for this issue has been merged into master.

from corda.