You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Reported by a user on a dual-stacked ISP connection. The ISP had an IPv6 outage, so the ISP router stopped doing v6 router advertisements, and over about 1h the various machines on the LAN lost their global IPv6 address. IPv4 connectivity continued uninterrupted.
However, tailscaled remained stuck on trying to speak IPv6 to control, log, derp, everything... And so had no connectivity. The logs (gracefully pulled out of journald by the user, since debug log collection was also borked) were full of bootstrap DNS attempts at connecting to IPv6 addresses, and failing with "network unreachable", since... yeah, nothing on the machine has any v6 state. Tailscale seemingly was only trying to connect over IPv6, never IPv4.
Restarting tailscaled made it correctly connect to everything over IPv4, so this seems to be an issue with a runtime transition from working v6 to no v6.
Machines affected, for debug log purposes: 100.67.182.67 was trying to ssh to 100.102.43.14 today at 18:56 Pacific. Tailscale on both machines was offline from this issue, so logs are going to be timestamped around 19:18 server time.
The text was updated successfully, but these errors were encountered:
Presumably this means tailscaled had resolved the hostnames to an IPv6 address, and never tried any of the IPv4 addresses. I imagine this is still the case now.
It is indeed still an issue, I'm running into it now. The strange thing is that we have dozens of supposedly identical machines and only one of them is having this issue.
What would be most useful is a tailscale bugreport from one of the identical systems which is not experiencing a problem and from the system which is. We can look at what their telemetry shows as being different.
We would need a bugreport from a system to be able to diagnose this.
I believe the original issue is no longer actionable, we wouldn't be able to reconstruct what happened now. If a similar symptom is seen again please open a new issue.
Reported by a user on a dual-stacked ISP connection. The ISP had an IPv6 outage, so the ISP router stopped doing v6 router advertisements, and over about 1h the various machines on the LAN lost their global IPv6 address. IPv4 connectivity continued uninterrupted.
However, tailscaled remained stuck on trying to speak IPv6 to control, log, derp, everything... And so had no connectivity. The logs (gracefully pulled out of journald by the user, since debug log collection was also borked) were full of bootstrap DNS attempts at connecting to IPv6 addresses, and failing with "network unreachable", since... yeah, nothing on the machine has any v6 state. Tailscale seemingly was only trying to connect over IPv6, never IPv4.
Restarting tailscaled made it correctly connect to everything over IPv4, so this seems to be an issue with a runtime transition from working v6 to no v6.
Machines affected, for debug log purposes: 100.67.182.67 was trying to ssh to 100.102.43.14 today at 18:56 Pacific. Tailscale on both machines was offline from this issue, so logs are going to be timestamped around 19:18 server time.
The text was updated successfully, but these errors were encountered: