Terragrunt execution order with apply --all

[maonat]

Hi,

I've implemented a workflow that is allowing us to enable "import mode" when a resource fail to create because it already exists and import it automatically if such flag is enabled.

While launching a nested directory as per below

❯ terragrunt list --tree
.
├── dir1
│   ├── dns-entry
│   ├── endpoint
│   ╰── server
╰── dir2
    ├── dns-entry
    ├── endpoint
    ╰── server

I get the following nice looking output

> terragrunt plan --all --non-interactive
....
00:00:00.000 INFO   The runner at . will be processed in the following order for command plan:
Group 1
- Unit ./dir1/server
- Unit ./dir2/server

Group 2
- Unit ./dir1/endpoint
- Unit ./dir2/endpoint

Group 3
- Unit ./dir1/dns-entry
- Unit ./dir2/dns-entry

In order to be able to import the entire path, I must launch up to 3 times (in this case) the apply --all because if it fails at the first group it will stop the execution of the apply for the second and third group.

---

Is it possible to obtain the details of how many groups I need to execute without having to launch plan, apply, validate or anything else that would use terraform?
Using terraform for such thing would take too much time because it would require to download the module every time and I want to skip that part if possible because seems that only terragrunt is required for such thing!

I tried looking into list but cannot find a way to get the same output which states exactly how many groups we are going to execute.

[yhakbar]

Hey @maonat ,

The deprecated output-module-groups is the old command that did this, but we're going to be removing support for it very soon, and we're going to make it so that Terragrunt no longer runs in "groups" anymore soon as well.

You can learn more about that here.

Basically, we're going to make it so that Terragrunt concurrency is more efficient, and runs happen as soon as they're no longer blocked by dependencies, instead of batching runs in groups like we are right now.

You can learn more about that here.

[maonat]

@yhakbar Do you have somewhere documented how it's the output going to look like in the GA version?
I understand that my best option will be to do the init and proceed with the apply afterward with a for-each loop where I repeat up to n times the apply-import function

[yhakbar]

Hey @maonat

I'm not sure what you mean with the loop import statement.

You can start using the functionality today with --experiment runner-pool

[maonat]

Is there a possibility to add in the output a total amount of recurrences that would occurr? Something sent in DEBUG so I can simply catch the queue "amount" making me able to repeat my script that apply (and if fails import the resource that failed to create only if already exists).
This would simplify me the work because I would not need to obtain that by myself.

BTW what is going to be the version that will have this as GA?

[yhakbar]

I'm sorry, I'm not fully understanding the question. If possible, please consider joining Terragrunt Office Hours. They take place every other week on Wednesdays at 10:30 ET (including today) in the Terragrunt Discord:
https://discord.gg/M3cmvFTXt4?event=1383440735310974997

We can chat about the issue that you're facing live, and help you work through it. Alternatively, you could create a simple fixture that reproduces the behavior you're trying to address, and ask how to achieve a certain outcome. That would make it so that we can more easily help you solve your problem directly.

We haven't set a firm version where this will be GA, but it should be in the next minor release or so (somewhere around v0.89.0).