-
Notifications
You must be signed in to change notification settings - Fork 1.7k
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
Support distributionPolicy
when creating regional instance group managers.
#1092
Support distributionPolicy
when creating regional instance group managers.
#1092
Conversation
@rosbo @danawillow @ndmckinley @paddycarver How do PRs get assigned to someone for review? There doesn't seem be any documentation explaining the process, and no activity at all for two weeks is sadface. :( |
Sorry @tobz, between the flu that's been going around, some travel, bugs we've needed to fix quickly, and other project work, this one seems to have slipped through the cracks. I'll take a look at it this afternoon :) |
@@ -145,6 +149,18 @@ func resourceComputeRegionInstanceGroupManager() *schema.Resource { | |||
}, | |||
}, | |||
}, | |||
|
|||
"distribution_policy": &schema.Schema{ |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
The thing that I'm not totally sold on for this approach is that with a name like "distribution_policy", it should really be a list of policies rather than a list of zones.
As much as the nesting is weird for just one field, it does make the resource more maintainable in the future (especially once we start generating resources). If the API maintainers decide to add another field to that block for example, we'd then have to deprecate this field and make it into a nested block, which is a bit of a pain.
Happy to hear a rebuttal though :)
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I'm not against such a change.
Like I mentioned in the opening comments, I made the choices I did to present a good UX to the end-user: ultimately, this attribute is simply for saying which zones in a region should be used.
I guess I'd be curious about how easily users could splat out a list of zones into such a list of nested blocks. I'm not sure I've seen that or done it myself.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Alright, so I ended up putting you suggestion into action. Now it's a list of objects, matching the structure of the API.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Sorry to go back and forth on this- your comment about being able to splat is a good one. I asked around a bit and the suggestion I got that I like is to just have a top-level list of strings called distribution_policy_zones
. That leaves room to deprecate it later in favor of a straight distribution_policy
object in case the API adds more things, but makes a good UX for the time being.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
You're killin' me, smalls! :)
Alright, I'll go with that approach.
@@ -227,6 +245,7 @@ func getManager(d *schema.ResourceData, meta interface{}) (*computeBeta.Instance | |||
|
|||
if err != nil { | |||
handleNotFoundError(err, d, fmt.Sprintf("Region Instance Manager %q", d.Get("name").(string))) | |||
return nil, err |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Good catch! This should just be return handleNotFoundError(...)
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
So, the signature doesn't actually support that specific code snippet.
What's curious, though, is that we'd be returning nil, nil
in the case of a not found error if we used the output of handleNotFoundError
. This feels totally wrong, but I'm willing to deal if that's simply the way things shook out in the provider.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
So what we want overall is to make sure that if the IGM isn't found when we try to read it, that we remove it from state. This is a totally valid case and shouldn't cause Terraform to error. So we don't want to return the actual error here, because then a not found will cause our terraform command to fail. Here's how we handle it for resource_compute_instance
:
https://github.com/terraform-providers/terraform-provider-google/blob/ab385ff5740b7a53ca6302d6c6214c273e178aa0/google/resource_compute_instance.go#L773-L776
https://github.com/terraform-providers/terraform-provider-google/blob/ab385ff5740b7a53ca6302d6c6214c273e178aa0/google/resource_compute_instance.go#L588-L603
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Makes sense. I'll take care of it.
Type: schema.TypeSet, | ||
Optional: true, | ||
ForceNew: true, | ||
Computed: true, |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I don't think this needs to be computed, does it?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I'd have to go back and test with/without it, but I believe it needs to be computed because while the API will let you pass in, say, us-central1-a
, whenever you retrieve the info for the rIGM, you will get back the full zone URL i.e. https://googlecloud.com/ur_project/zones/us-central1-a
, and then TF would freak out thinking there needed to be an update.
Let me try to retest that one. I'm also open to ideas on generating the full one URL behind-the-scenes, but it seemed like a bunch of extra work at the time, and I wasn't really familiar with a way to easily nab the current project, etc, to build it.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Nah, that's what the diffsuppress is for. Computed means that if it isn't set, the API will provide a value (so don't show a diff when the config doesn't have a value for that field).
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
So, I went back. This is actually required.
If we use the beta API version, the API actually fills in the distribution policy for you, based on -- what appears to be -- the zones the rIGM seeds itself with during creation.
} | ||
} | ||
|
||
return schema.NewSet(schema.HashSchema(&schema.Schema{ |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I think just return zones
will work (and then update the return type)
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
FWIW, I tried this, to no avail.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I just uploaded a commit where this seems to work. Let me know what you think!
@danawillow I think this is ready for another pass. |
This approach lets us more simply allow a list of zones to use, while providing a deprecation path for implementing the distribution policy field more holistically, avoiding backwards-incompatible changes.
@danawillow Alright, I think we've arrived. :P |
FYI I just pushed a typo fix and the change in flatten that I was talking about. If that seems all right to you, I'm ready to merge this :) |
Thanks for the fix-ups! I now see what you meant by changing the flatten method. :) Tests pass; LGTM. |
For my own edification, when are releases usually cut? |
We aim for every 2 weeks, but sometimes things come up and it stretches a bit. If all goes according to plan, we'll do a release on Monday, but no promises. |
Late response: yeah, not in a rush, just curious. :) Thanks again! |
…nagers. (hashicorp#1092) * Support `distributionPolicy` when creating regional instance group managers. * Better match the API structure of distributionPolicy. * Switch to "distribution_policy_zones". This approach lets us more simply allow a list of zones to use, while providing a deprecation path for implementing the distribution policy field more holistically, avoiding backwards-incompatible changes. * fix typo * use slice instead of Set for flattenDP
Signed-off-by: Modular Magician <[email protected]>
I'm going to lock this issue because it has been closed for 30 days ⏳. This helps our maintainers find and focus on the active issues. If you feel this issue should be reopened, we encourage creating a new issue linking back to this one for added context. If you feel I made an error 🤖 🙉 , please reach out to my human friends 👉 [email protected]. Thanks! |
This addresses #549.
This one is sort of a doozy because it's both an un-updatable attribute and beta feature. The format required for the zones to be specified in was also a little tricky. Definitely looking for thoughts on how my approach could be better. I chose what I thought was the best user experience -- users specifying the "naked" zone -- and augmenting it behind-the-scenes.