-
Notifications
You must be signed in to change notification settings - Fork 135
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[BUG] chart version 3.0.0 throws "Your DNS resolver at 10.43.0.10 isn't doing DNSSEC validation; Please use another resolver or enable unbound." #144
Comments
I'm running it without problems (on bare metal with rancher2). Possible solutions are described here: Can you try to set for the admin pot:
But I'm not sure it works since the default k8s resolver is still in /etc/rsolve.conf. |
I'm hitting this issue too, on a new install. Using Helm to AKS. Tried the DNS setting above, it does not have an effect when used with If you disable I have tried adding the below to the admin container but without success.
I am using Core-DNS within the cluster. |
Indeed, tried
and also other dns configurations, but then the local pods can't be resolved, and it seems its depending on redis. I'm not seem to fix it either by setting the DEFER_ON_TLS_ERROR env value to false. |
Please see current discussion in the matrix chat. Most important result to me is:
You should check that your upstream dns (that configured on your host) has properly working DNSSec: https://dnsinstitute.com/documentation/dnssec-guide/ch03s02.html. You should also check which upstream DNS your CoreDNS actually uses. |
Hey @micw, thanks for digging into this. I started doing some tests. I created a new pod, based on an ubuntu image.
It contains the DNSSEC OK (do) flag indicating the recursive server is DNSSEC-aware.
"forward . /etc/resolv.conf" means it forces all non-cluster DNS lookups to go through a specific nameserver at /etc/resolv.conf |
Another test:
Maybe the mailu-v1.9 code regarding the check whether my DNS server supports DNSSEC is not 100% correct? |
Another test, this time I'm making use of the admin-v1.9 container itself. I've installed dig, using 'apk', since the admin container seem to be based on the alpine image:
Conclusion:
|
I was facing the same issue yesterday, and although all the checks I tested worked, with the right flags, in dig I could not get it to work, so I deployed unbound on a pod and it is all running great. The default unbound config from mailu will not work since coredns does not support dnssec, so here is what I used in a configmap mapping into to the unbond.conf: apiVersion: v1 |
@GreatWhiteDan thank you. I'm open to add this to the chart. Can you explain how a pod can use this and still resolve the internal addresses? |
Great to hear @GreatWhiteDan. I'm also curious how the admin-pod makes use of the unbound container. And you're saying 'coredns does not support dnssec', can you elaborate on this? |
Sure, you can add this to your deployment for the admin pod:
|
@kooskaspers I'm afraid you're running the dig query wrong. Mailu requires a DNSSEC resolver that's doing validation as neither the admin app nor postfix can do it on their own. The AD flag says that the response data has been authenticated: that's what the admin container checks (see https://github.com/Mailu/Mailu/blob/2c1d1e995d0ff7e52684c67d72265ff3b4439341/core/admin/start.py#L42 and https://datatracker.ietf.org/doc/html/rfc6840#section-5.7 if you're interested in the specifics). Being DNSSEC "aware" is not enough. Please post the output of those three commands:
None of the snippets you've pasted so far feature the |
@GreatWhiteDan Please be mindful that with the configuration you have posted above, ACLs are disabled and you may end up providing recursive DNS resolution services to the WWW (which will most likely make you become a DDoS relay).
|
Well first of all, thank you for bringing me a bit up to speed regarding the topic; appreciated.
Here we go:
The AD flag is indeed missing when quering with the local cluster resolver, CoreDNS. |
I was expecting I think that Mailu behaves like expected: it warns you that something is off. |
Yep, I got it working now. Enabling DNSSEC in pihole, even though I route the DNS requests through my router to this pihole server, has resolved the problem! |
Note to anyone who comes here due to a critical outage and you just need to get things back up and working. You can always use the previous version 0.2.0 for now so you can get this whole DNSSEC thing working when you have a little more time. |
another note for anyone using k3s and getting this: thanks to everyone that helped analyzing this! |
@kooskaspers I had to enable the |
Thanks mate. Came back to my own topic here, after having troubles with the same error I receiver earlier! |
@kooskaspers I don't know what the workaround is (as in, how it works), but you can check out the related PR or the complete 1.9.10..1.9.20 diff |
The workaround is to set the Both of those commands should return responses with the
|
Describe the bug
Upgraded to chart version 3.0.0 (mailu 1.9.10). I've updated my values.yaml with new variables. After upgrading, the admin throws error:
Your DNS resolver at 10.43.0.10 isn't doing DNSSEC validation; Please use another resolver or enable unbound
I'm using cloudflared as a dns over https dns service. I've switched to google (8.8.8.8) after receiving this error but to no avail. I'm curious whether more people on k8s have encountered this issues on chart version 3.0.0. Anyone succesfully upgrade to 3.0.0 as well?
Environment
Additional context
The text was updated successfully, but these errors were encountered: