-
Notifications
You must be signed in to change notification settings - Fork 50
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
Updating tags on AKS cluster attempts to "replace" it, but ends up deleting it completely #182
Comments
To be clear, although I think it's odd that a tag change would require tearing down a cluster in the first place, if that is required.. so be it.. but I'd expect that there still is a cluster (a new one) that is left over at the end over the operation.. there isn't! There is NO CLUSTER at all! Sorry if I wasn't clear. |
Yes - this doesn't look right - we'll look into it. |
Updating a tag should not cause a replace. The docs say it's a |
I looked into this. It is actually another case similar to pulumi/pulumi-aws#451. The ultimate offender is In the TF schema it looks like:
This value is computed, and gets populated with the value If I add
There are three issues to follow up on:
|
This apparently does not reproduce with Terraform, so we'll need to look at what fix is appropriate on the Pulumi side.
|
Opened pulumi/pulumi#2453 on (1) above. |
I wasn't able to reproduce this aspect of the original bug report. Even when I do cause a replace to be triggered - it does successfully create a new cluster and delete the old cluster.
This is also surprisingly short for a replace to have actually happened. For me this took |
@lukehoban try setting the {name} parameter on the cluster properties, see if that triggers the "delete" aspect.. that's how I had it. This is my personal account, I filed the bug report from work account (I was forced to create a separate work one...) |
Here's the code:
|
Okay - I can reproduce the disappearing cluster issue - and it is indeed tied to specifying What happens is:
The root of this is (3), which I believe is being changed in the A workaround now (which will be necessary to do the replacement at all after the |
Opened 3 issues to track various pieces of this:
I'll close this one itself out and we'll track in those other three issue. |
FWIW, I cannot reproduce this with the latest bits. Looking through the history, it looks like this commit changed the relevant field from an optional forcenew with a default to an optional computed forcenew. If I remove the computed annotation from the schema, I can reproduce this issue. |
Found this out while doing some dev work. Every time we deploy an AKS cluster, we have a tag of "created" which is the current timestamp. So while iterating during development, I found out that if you redeploy the stack (which includes the AKS cluster, which includes a new tag on that cluster) the cluster will attempt to replace itself (for a tag?) but it ends up seeming to "create" a new one (of the same name) which I think is simply an update.. and then "deletes the old one" of the same name.. which means you end up with a deleted cluster.
It's gone! Is this pulumi or TF provider?
The text was updated successfully, but these errors were encountered: