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
A reboot will reset the upstream DNS to the default on reboot.
I have not been able to figure out how to make this setting persist reboot. Disabling DNS is off the beaten path, and there's some weird systemd/resolved/cloud-init network configuration stuff going on that I couldn't figure out.
On TPP, running resolvectl dns eth0 "" removes the default upstream DNS resolution, leaving the host with only the static IPs in /etc/hosts, which is what we want.
However, in the DO test backend, and in the local lxd VM, this does not replace the default DNS, just adds to it.
It may be something to do with systemd version, or possibly differences in the user-data or vendor-data passed to cloud-init that the various VMs were provisioned with.
Either way, I can't figure out how to get it to survive reboot. I suspect the answer lies in modifying the cloud-init user-data, perhaps. But I've run out of time, and need to park this.
The text was updated successfully, but these errors were encountered:
Couldn't figure out how to make cloud-init reliably stop managing
/etc/hosts, so instead modify the cloud-init template and let it do it.
The install script tests this has worked, or exits
Tried to figure out how to reliably disable the default upstream DNS.
`resolvectl dns eth0 ""` works in TPP, but does not work in test backend
or in the local lxd test VMs. Ran out of time/ideas to figure out why,
so added a per-backend post-install hook to disable it in TPP, more
details in #240
The result of this is that our TPP static DNS config should now not be
wiped by a reboot. We'll still need to re-run just install to disable
upstread DNS again, however.
A reboot will reset the upstream DNS to the default on reboot.
I have not been able to figure out how to make this setting persist reboot. Disabling DNS is off the beaten path, and there's some weird systemd/resolved/cloud-init network configuration stuff going on that I couldn't figure out.
On TPP, running
resolvectl dns eth0 ""
removes the default upstream DNS resolution, leaving the host with only the static IPs in /etc/hosts, which is what we want.However, in the DO test backend, and in the local lxd VM, this does not replace the default DNS, just adds to it.
It may be something to do with systemd version, or possibly differences in the user-data or vendor-data passed to cloud-init that the various VMs were provisioned with.
Either way, I can't figure out how to get it to survive reboot. I suspect the answer lies in modifying the cloud-init user-data, perhaps. But I've run out of time, and need to park this.
The text was updated successfully, but these errors were encountered: