Terraform Error: Failed to query available provider packages
chris2k20 opened this issue · 4 comments
First thanks for your work on this project ! :)
When running make infrastructure BOOTSTRAP=true
I get the following error okd.log
:
...
Error: Failed to query available provider packages
Could not retrieve the list of available versions for provider
hashicorp/hcloud: provider registry registry.terraform.io does not have a
provider named registry.terraform.io/hashicorp/hcloud
Error: Failed to query available provider packages
Could not retrieve the list of available versions for provider
hashicorp/cloudflare: provider registry registry.terraform.io does not have a
provider named registry.terraform.io/hashicorp/cloudflare
make: *** [Makefile:86: infrastructure] Error 1
I already performed some actions on the terraform versions.tf but without success..
Hi @chris2k20,
this is caused by the new terraform version and should be fixed in the feature branch upgrade-to-4.7.x. Well, enable Renovate with auto merge wasn't the best idea... :)
https://github.com/slauger/hcloud-okd4/tree/feat/upgrade-to-4.7.x
There are three major problems that I am currently working on in this branch:
- Something changed in the cluster bootstrap with OKD 4.7. I did not had luck installing OKD 4.7 on hcloud yet. Your help is appreciated. :(
- The default resolvers from hcloud have a very strange caching mechanism and ignore the TTL from Cloudflare. A newly created record (e. g.
api-int.cluster-id.basedomain.tld
) is resolvable after about 60 minutes. This record is required for the inital boot of all master/worker nodes -> we need somehow override the dns servers provided by dhcp and switch to custom dns servers / cloudflare (during the first boot!) - RPC Bind is reachable from the public internet by default (this could be fixed with the new firewall beta feature which is available since today -> https://www.hetzner.com/news/03-21-firewall-beta/)
The poor I/O performance of the hcloud (which could cause etcd issues) still remains unsolved.
Should this be closed?
The problems still persist. I think we're leaving the issue open.