Re: v3 probe stuck with BADMD-APP
Hello, I have had recently the same problem like Radu described. The probe wasn't connecting with this message since 2025-11-07 09:45:52, so I have taken this probe to different ASN, but it didn't help. Today, after configuring the DNS servers in the original ASN to respond to queries coming from the probe’s temporary network, the probe was finally able to reconnect. It looks like the issue with DNS setting was a problem here. The probe was initially set to use static IP and IPv6 addresses. After changing a few weeks ago to dynamic ones it's still stuck with old DNS server settings. It's really hard to fix the USB issue with the probe using static addresses. The only working way was taking it to the network with working DHCP and/or IPv6 auto-configuration including RDNSS, plugging it without USB, waiting several minutes, plugging USB stick, waiting till it reboots, disconnecting from the network, and connecting back to the network for which static addresses were configured. Of course, if there were DHCP or RDNSS working in the original network, the troubleshooting would be far easier. Anyway, I have two requests for the RIPE Atlas operations team: - please ensure that DNS settings automatically switch to dynamic when changing a probe’s addressing mode from static to dynamic, - please allow users to change a probe’s addressing mode from static to dynamic even when the probe is currently offline. Cheers -- Marek Zarychta
Hello Marek, On 18/11/2025 09:46, Marek Zarychta via ripe-atlas wrote:
Anyway, I have two requests for the RIPE Atlas operations team: - please ensure that DNS settings automatically switch to dynamic when changing a probe’s addressing mode from static to dynamic
We have added this in the UI recently [0]: when probes are switched from static to dynamic, DNS servers are also wiped. As this is a UI change, we have not modified the networking settings for previously configured probes. That is, if your probe was set as dynamic, but had DNS servers set, those stayed.
- please allow users to change a probe’s addressing mode from static to dynamic even when the probe is currently offline.
We cannot change a hardware probe's network configuration remotely while it's disconnected from the back end. To avoid having a discrepancy between a potential "to be applied" network configuration in the UI, and the probe's actual configuration, we don't currently allow for doing this through the UI. The probe should be able to be rescued, though: the current probe firmware should do fail-over to dynamic IPv4 if static IPv4 fails. Our current check for "static IPv4 failing" is pinging the static IPv4 gateway, and seeing if 'arp -n $IPV4_GW' returns a mac address or not. [1,2] This approach has a few downsides, notably the lack of testing for IPv6, and the lack of testing for connectivity through that gateway (e.g. to RIPE Atlas servers), but it has been in-place ever since we added static config support. Note that there been internal discussions and ideas on how to replace this with a better test. -- Regards, Trix Taiclet Embedded Systems Engineer @ RIPE Atlas RIPE NCC [0]: https://atlas.ripe.net/docs/releases/website_index/2025-10-09 [1]: https://github.com/RIPE-NCC/ripe-atlas-software-probe/blob/1fe2c6a381a02aece... [2]: https://github.com/RIPE-NCC/ripe-atlas-software-probe/blob/1fe2c6a381a02aece...
participants (2)
-
Marek Zarychta -
Trix Taiclet