We have been using aaa.bbb.local for our network hostnames when we move to production. In testing we had all of our RIOs using fixed IPs for the broker. Now that we are switching to production and moving to hostnames, we are having issues getting the hostnames to resolve with Ping from the RIO and making a connection to the broker. When we run a Nslookup from the RIO, it resolves fine. We can put a windows PC on the same port and everything resolves and works fine. My IT manager is starting to suspect that Ping on the RIO (linux) and the connection to the broker are viewing the .local hostname as a multicast address so it won’t resolve. Has anyone else run into this issue?
The Internet is littered with ‘its fine’ and ‘no, dont use .local’…
That said, clearly you are finding an issue with it on your network.
I think if your IT DNS is handing out .local hostnames the issue could be there. There is nothing really unique about the Linux on RIO (or EPIC), so it just comes down to how you are reserving the hostnames per MAC for each RIO in the DNS setup.
Thanks for the info. Our former MSP stuck us with .local and it had been working until it didn’t. Our new staff IT manager knew it might be a problem when he came on board, just not where.
We are planning to replace .local with something else as soon as we can. (after we fix the other problems they left us.) We have been using .local for all the RIOs and EPICs without issue and fixed IPs for the brokers. We already have 100 devices out in the shop, so it will be a long weekend when we swap everything over.
It’s not yet official, but ICANN is probably going to reserve “.internal” for purposes such as this. Your home’s internet connection could soon be called .internal - The Verge