Closed toxuin closed 4 years ago
Sorry to hear that :( if you revert to the earlier version do they come back?
Yes, that is what I did for now.
Interesting, I'll have a look and see what I can see.
What about 1.7? Just trying to narrow down the code diff.
Version 0.1.7 seems to work fine!
I'm guessing it has something to do with k8s version checking to determine new endpoints or related. I use the service plugin without issue however on a 1.15 cluster without issue so that's a bit strange. I may have you send me the output of the same resource using the 2 different endpoints and see what we get.
You mean ingress resources /apis/extensions/v1beta1/ingresses
vs /apis/extensions/v1beta1/ingresses
?
Then here you go 😄
Update: silly me, I have accessed both resources with a cluster admin account. Impersonating service account shows different picture: 403 on /apis/networking.k8s.io/v1beta1/ingresses
. I should've noticed it sooner! D'oh!
Yeah nice find. I didn't see it because I run out of cluster with admin config :(
0.1.8 with change in ClusterRole does fix most of the error messages, but one persists:
2019-11-20T23:23:03+00:00 plugin (pfsense-dns-ingresses): failed saving dnsmasq config: Read timed out after 10 seconds (1000)
11/20/2019 4:23:03 PM 2019-11-20T23:23:03+00:00 plugin (pfsense-dns-ingresses): failed update/reload: Read timed out after 10 seconds (1000)
Just tried to delete a service with a hostname and had another one:
PHP Warning: Illegal string offset 'host' in phar:///usr/local/bin/kubernetes-pfsense-controller/src/KubernetesPfSenseController/Plugin/Utils.php on line 136
11/20/2019 4:30:22 PM PHP Warning: Illegal string offset 'host' in phar:///usr/local/bin/kubernetes-pfsense-controller/src/KubernetesPfSenseController/Plugin/Utils.php on line 136
Do you use dnsmasq?
I use unbound
. It actually worked (delete service with hostname = remove alias in unbound), despite all the errors in logs. I have noticed that recent update had some more logging functionality, so maybe it was always like that, just never visible? 😄
Turn off dnsmasq in your configmap.
That clear everything up?
Seems to be the case!
Using version 0.1.8, just updated from 0.1.5 and all my cluster domain names are gone from DNS resolver settings. I am seeing warnings and errors in logs of the controller:
ConfigMap
kubernetes-pfsense-controller-store
has{"managed_hosts":[]}
for bothpfsense-dns-ingresses
as well aspfsense-dns-services
.Running kubernetes version 1.15.5