Core deployment errors - Error: reading blob properties for Storage Account

[jonnyry]

Deploying a fresh TRE from 2 commits behind main (1ffb09b)...

Experienced storage account failures on two seperate deployments:

[tim-p-allen]

could be an issue with the terraform providers?

[jonnyry]

Just pulled your tf provider update and giving it a try now.

[jonnyry]

Reployed fine with the tf provider update - I'll close this off.

[jonnyry]

Unfortuntely just happened again on the latest codebase (including the tf provider update):

[tim-p-allen]

how weird, and the storage account doesn't exist in the portal?

[jonnyry]

I have already destroyed that environment already, but I've just seen it happen again and I've checked and the storage account does exist:

[jonnyry]

Looks like it could be this: hashicorp/terraform-provider-azurerm#22992

[jonnyry]

And for the issue to be resolved, wait for the following PR: hashicorp/terraform-provider-azurerm#26218

[jonnyry]

still occurring on almost every deployment:

[marrobi]

Are you using the same TRE ID between deployments? Wonder if some sort of DNS caching going on?

[jonnyry]

I am yes. I think it is to do with DNS caching.

However not sure why the issue has only started occuring this week, as been reusing TRE IDs for a long time.

[marrobi]

Are these jobs running on a github agent? Is it a GitHub hosted one or an agent you manage?

[jonnyry]

Just a standard GitHub Actions Runner - not one that we manage.

[marrobi]

Hmm, we never really deploy, destroy, and deploy with the same ID. Wonder if you left it a couple of days before reusing the same ID.

Might be need to create an issue in the Terraform provider repository/wait for the one you mentioned, not sure its an Azure TRE issue, or something we are going to be able to resolve.

[jonnyry]

It's intermittent - I think it does depend on how long it was since the last deploy.

Curiously for the issue I was having this morning, I deleted the storage account that was causing the 404, re-ran the build and it completed sucessfully.

I'll hopefully manage to work around it for now and keep an eye on release of the terraform PR above.

[marrobi]

Could be fixed by https://github.com/hashicorp/terraform-provider-azurerm/releases/tag/v3.113.0

[jonnyry]

Strangely not seen it for a couple of weeks - it went away by itself. If it reoccurs, I'll try bumping the terraform azurerm version.

[marrobi]

Going to close this given not senit again.

[jonnyry]

This has occured 6 times in the last week - when attempting to deploy a TRE instance using the same name as that of a recently deleted TRE instace.

Going to try bumping TF provider in the core from v3.112.0 to v3.117.0 (incorporating a fix for this issue) to see if it helps.

[jonnyry]

OK three deployments using v3.117.0 have succeeded.