sync_ping issue

I'm using resource_discovery in an environment where I want to bring nodes up and down regularly.

The rd_util:sync_ping/2 function waits until the nodes(known) on a remote matches in number the nodes(known) on this node. It fails if this doesn't happen within timeout.

If I have a 'contact' node, and two nodes 'foo' and 'bar' then this sequence will break it:

1. Bring up 'contact'.
2. Bring up 'foo' and 'bar'
3. Bring down 'foo' and 'bar'
4. Bring up 'foo'.

After step 4, the 'contact' node has 3 nodes known, and the 'foo' node has 2 nodes known.

I am wondering why the cardinality of 'known' nodes is important?

I would suggest simply using nodes() in line 49, not nodes(known).

[martinjlogan]

ethertricity nodes(known) will cause this problem if the nodes are brought up and down rapidly. The reasoning for nodes(known) is historical and had to due with the production system that originally used resource_discovery. I agree that this should be removed. If you have a patch please send a pull request and I will take care of it. If not, I will make the change myself when I am back from vacation.