Broadcast PING

[WilliamM]

In my opinion, the Broadcast PING feature in Portal 2.1 BETA is not a very effective way of re-discovering nodes. Giving ample time between broadcasts, it took me 10 pings to discover 36 of 40 known, operating nodes. At first, I'm finding about 5 nodes with each ping; towards the end, though, it becomes almost helpless.

With 40 nodes, it is faster and more reliable to select all 40 in the "All Nodes" node view, and choose refresh - allowing Portal to systematically check each node's status. One cycle of this finds 40/40 active nodes.

Obviously, Broadcast PING is still necessary to discover nodes for the FIRST time. And once the number of registered nodes is greater, it will become impractical to refresh the entire "All Nodes" list. Because of this, I would like to see a Broadcast PING that is better at handling large quantities of nodes in the future.

[WilliamM]

Node Watcher works well for removing nodes that are no longer active from the "Active Nodes" list. Refreshing all of the "Active Nodes" also works, and takes about as much time, but generates a big back-up in Queued RPCs.

[mgenti]

I would like to see a Broadcast PING that is better at handling large quantities of nodes in the future.

We plan on adding a feature in version 2.1 of Portal that should address this issue. At this time we do not know when this will be included in a beta release of 2.1

[WilliamM]

Could you tell us when you do know? We are waiting on this. Otherwise, we may have to write this method ourselves.