@joshuef (IIRC) you mentioned that the node has an improved checking for the port # already in use.
Well since the upgrade I’ve had the node throw this too often. Be it with safenode-manager starting the node or my script starting the node.
Today saw it throw that the metrics port I specified was already in use. And so I was ready for this and did a
ss -l | grep of the port number
and got zero results.
With the number of times the error occurred when there should have been zero possibility and now catching it and checking for port in use and finding node, I suspect there is some issue in the port in use detection.
Hey @neo!
You might want to make sure your node registry is clean. If any of the ports is already set inside the node registry, then it’ll throw an error, even if the service from the registry is not running.
Maybe something to tweak later.
These are happening from running the node directly as well. And yes I start them fresh. Well usually its worse when there are already nodes running, but that is more like randomness, the first node has the small chance of it happening, but starting 20 has 20 times the chance of one seeing the error. Its enough that I have stopped using node-manager since sometimes when it happens on the listening port the node-manager stops (IIRC)
Doing the scripting myself I can retry running the node on those ports after starting the others. Node manager seems to fail on that node no matter how many times you try to start it