#terraform-tool

Hi gents, what's the current best practice for reusing configuration in multiple environments?

anyone have any examples?

rbowlby: var lookups with different state files

is it expected that, when using a terraform variable (tfvars) containing a variable expansion such as "${file(<include>)}", with the template provider, the rendered file will contain the contents of the file <include> for the expanded variable?

hi! I'm getting this error! Parameter encrypted is invalid. You cannot specify the encrypted flag if specifying a snapshot id in a block device mapping when creating a new instance with root_block and a ebs_block

there is this issue https://github.com/hashicorp/terraform/issues/1862, but I cannot workaround it, any idea?

as with the issue, I'm using a ebs_block with a snapshot_id

this is my log http://hastebin.com/sokabizura.zsh

anybody?? mitchellh phinze ?? :)

question, I have changed user_data on a aws_launch_configuration resource, plan show me as -/+ operation

question is, it will impact on instances already running

will it?

QiQe: user_data is only applied when a node is instantiated

QiQe: so the nodes which are already running won't be affected directly (unless they re-run cloudinit for some reason).

thanks very much phrawzty

pretty please someone?

blackjid: I have compiled terraform 0.5.2 with the whole encryption for ebs parts ripped out: https://repo.gotapaas.eu/files/terraform.tar.bz2

use at your own risk though

ahp, thanks! herself I'll try it

what shoould I do with that?? umcompressed and there is a bunch of terraform-* files.... put I cannot execute them..

herself?

those are terraform binaries

you must hve installed terraform before somehow

thnks..

hey everyone!

whilst working on the Azure support, me and svanharmelen seem to have run into a blocker with regards to the creation of networking elements.

long story short; the Azure API does not allow for operations on networking resources to be done concurrently in a safe manner...

weirdly enough though; I can vouch that adding a mutex upon the C_UD operations of the affected resources seems to be able to solve the problem.

could someone with enough of a grasp on Terraform's internal provider handling methods please shed some light on all of this?

namely, does Terraform run a single provider for all the resources on the respective platform; or does it start up a provider for EACH resource it requires from that platform?

phinze and/or mitchellh please correct me if I'm wrong, but isn't this exactly what it is: https://github.com/hashicorp/terraform/pull/205...

Of course the source code also shows this, but I understand it can be a little overwhelming to just dive in there. Just for fun you could add a small log line to the `Provider()` func of the Azure provider to show you during execution how often it is called.

even better you could change the `Provider()` func a little so it doesn't directly return the `&schema.Provider{…}` but instead assign it to a variable which you can then log the pointer address from…

that will also show you at runtime that it will be called multiple times having a different address each time

so something like:

I have no problem being wrong; my problem is why does it work by adding a mutex?

test := &schema.Provider{…}

log.Printf("ADDR: %p", test")

return test

not sure…