We use extra large images for deploying and we have a limit of 20 cores
per project. So in order to save the no. of cores used we comment ood-
related code because it is not required for testing.

Collapse compute node and ood ops provisioners into a single
null_resource to avoid a race condition when the ansible tasks
run in parallel.  This lead to inconsistent deploys where
the slurm configuration was not working after deploy.

Putting the provisioners in a single resource ensures they run in
series.  This comes at the cost of a loss of deploy flexibility.
Now ood and compute nodes ops steps will always run on changes rather
than selectively.  Since this feature wasn't fully functional, it's
loss is not significant.

There is no call into ood by terraform so trigger it's ops
steps via an ansible-playbook during the user-data script.

Add file provisioner to sync the submodule state.

Eesaan Atluri authored 5 years ago and John-Paul Robinson committed 5 years ago

Packer image build will fail to git clone any repos residing inside
internal gitlab. So we define the internal DNS IPs in the network
module while creating the initial infrastructure for packer img builds

Bo-Chun Chen authored 5 years ago and John-Paul Robinson committed 5 years ago

To avoid error when there's two images with same name
 use data source to get the latest image to boot up compute nodes

Bo-Chun Chen authored 5 years ago and John-Paul Robinson committed 5 years ago

Changing ohpc id and compute node id will trigger it
Will run node registration on ohpc node.

Bo-Chun Chen authored 5 years ago and John-Paul Robinson committed 5 years ago

Accessible by calling module.create-ohpc-instance.id
 will return a string

Bo-Chun Chen authored 5 years ago and John-Paul Robinson committed 5 years ago

Accessible by calling module.nodes.network
 will return a map which key is node name and its network info

Bo-Chun Chen authored 5 years ago and John-Paul Robinson committed 5 years ago

Accessible by calling module.nodes.id
 will return a list of compute node id
Mainly using for null-resource trigger

Bo-Chun Chen authored 5 years ago and John-Paul Robinson committed 5 years ago

Modify warewulf config file with compute node info

Bo-Chun Chen authored 5 years ago and John-Paul Robinson committed 5 years ago

c0 to compute

Packer image build will fail to git clone any repos residing inside
internal gitlab. So we define the internal DNS IPs in the network
module while creating the initial infrastructure for packer img builds

Make modular

See merge request rrand11/terraform-openstack!27

# Conflicts:
#   main.tf
#   vars.tf