make domain optional #1803
make domain optional #1803
Conversation
| "\n\n 🪴 Great! Now you need to provide a valid domain name (i.e. the URL) to access your Nebri instance. " | ||
| "This should be a domain that you own.\n\n" | ||
| "\n\n 🪴 Great! Now you can provide a valid domain name (i.e. the URL) to access your Nebari instance. " | ||
| "This should be a domain that you own. FIXME Default is the IP of the load balancer\n\n" |
There was a problem hiding this comment.
Need to put something else here.
|
|
||
| print("#" * 60) | ||
| print(repr(inputs.domain_name)) |
There was a problem hiding this comment.
Oops
| print("#" * 60) | |
| print(repr(inputs.domain_name)) |
| print(inputs.dict()) | ||
|
|
There was a problem hiding this comment.
Oops
| print(inputs.dict()) |
|
Hey @pmeier, what is the status of this PR? |
|
Hi @pmeier, regarding what we discussed during the last community meeting, for the destroy part, if you would like the load balancer address (or any information from the resources) to be exposed we can follow a similar approach that was done for the address test during the deploy, right now, we are getting the LB IP from terraform here we then use this function to catch all the outputs (that terraform saves under a json file), so that its consumed here under the stage_output objects. Now, for destroy, we have this function gather_stage_outputs that will refresh all resources, and grab the needed outputs (including the IP mentioned), so you could use it in the destroy script as well in a similar fashion as the test function discussed above. |
|
Also, it might be interesting to run the Kubernetes CI on this PR as well, so that we can see the error message as well. I am a bit curious to see where terraform might be complaining about the address, as this should only be failing in the schema validation |
pmeier
added
status: in progress 🏗
|
Closing since implemented in #1833 |
Reference Issues or PRs
Closes #1707.
What does this implement/fix?
Testing
Any other comments?
nebari init(with or without--guided-init) andnebari deploywork fine. However we get problems for everything else since we never write back the IP of the load balancer that we find during deploy. I think we have two options here:domainfield a sentinel. During the verification stage, we could inspect the cluster, if available, and extract the domain from there.