Terraform-Elastifile-GCP

Terraform to create, configure and deploy a Elastifile Cloud Filesystem (ECFS) cluster in Google Compute (GCE)

Note:

Follow the Elastifile Cloud Deployment GCP Installation Guide to make sure ECFS can be successfully deployed in GCE before using this.

Use:

1. Create password.txt file with a password to use for eManage (.gitignore skips this file)
2. Specify configuration variables in terraform.tfvars:

• TEMPLATE_TYPE = small, medium, standard, custom. Only use custom in consultation with Elastifile support
• COMPANY_NAME = Name of the company that uses this cluster
• CONTACT_PERSON_NAME = Contact person name
• EMAIL_ADDRESS = Email address of the contact person
• NUM_OF_VMS = Number of ECFS virtual controllers, 3 minimum for small/medium, 6 minimum for standard
• LB_TYPE = none, dns, elastifile, google
• DISK_TYPE = local, ssd, or hdd. Only applies to custom templates
• DISK_CONFIG = [disks per vm]_[disk size in GB] example: "8_375" will create 8, 375GB disks. Only applies to custom templates
• VM_CONFIG = [cpu cores per vm]_[ram per vm] example "20_128" will create 20 CPU, 128GB RAM VMs. Default: "4_42" Only applies to custom templates
• CLUSTER_NAME = Name for ECFS service, no longer than
• EMS_ZONE = EMS Zone
• PROJECT = Project name
• SUBNETWORK = Subnetwork to use. default or full path to use specific/custom project or shared vpc subnetwork eg projects/support-team-172804/regions/us-west1/subnetworks/andrew-shared-vpc-network-subnet
• IMAGE = EMS image name
• CREDENTIALS = path to service account credentials .json file if not using
• SERVICE_EMAIL = service account email address
• USE_PUBLIC_IP = true/false. true for creating ems with public IP. false for creating ems with private IP only.
• DEPLOYMENT_TYPE = single, dual, multizone
• NODES_ZONES = list of the zones for the nodes
• SETUP_COMPLETE = true or false
• ILM = true or false
• AsyncDR = true or false
• LB_VIP = IP Address outside the subnet or auto - for elastifile LB only
• DATA_CONTAINER = data container name
• EMS_ONLY = true or false
• KMS_KEY = the encrypted customer managed keys for the ems boot disk

3. Run 'terraform init' then 'terraform apply'

Components:

google_ecfs.tf Main terraform configuration file.

create_vheads.sh Bash script to configure Elastifile eManage (EMS) Server via Elastifile REST API. EMS will deploy cluster of ECFS virtual controllers (vheads). Called as null_provider from google_ecfs.tf

Note: REST calls are HTTPS (443) to the public IP of EMS. Ensure GCP project firewall rules allow 443 (ingress) from wherever this Terraform template is run.

destroy_vheads.sh Bash script to query and delete multiple GCE instances and network resources simultaneously. Called as null_provider destroy from google_ecfs.tf

add_vheads.sh Bash script to query and add multiple GCE instances and network resources simultaneously. Called as script from update_vheads.sh

remove_vheads.sh Bash script to query and remove multiple GCE instances and network resources simultaneously. Called as script from update_vheads.sh

update_vheads.sh Bash script to query, add/remove multiple GCE instances, update the google iLB and network resources simultaneously. Called as null_provider from google_ecfs.tf

create_google_ilb.sh Bash script to create all the neccesary resources for the google iLB simultaneously. Called as null_provider from google_ecfs.tf

update_google_ilb.sh Bash script to query and change the google iLB configurations after adding/removing nodes. Called as script from update_vheads.sh

destroy_google_ilb.sh Bash script to query and delete all the google iLB resources simultaneously. Called as null_provider destroy from google_ecfs.tf

password.txt Plaintext file with EMS password

Troubleshooting:

** .log files* Output of all scripts and REST commands

/elastifile/log/ Log directory from EMS

Known Issues:

Custom template configurations are not officially supported by Elastifile. Shared VPC configuration is partially supported, and will cot configure the Elastifile LB.

This version supports Elastifile Ver 3.x with the following:

• Single replication for SSD PD device configurations.
• Dual replication for all configurations
• Multizone cluster, needs 3 zones in the same region.
• Adding and removing nodes from a live cluster, by changing the SETUP_COMPLETE to true, and modifing the NUM_OF_VMS to the requested number.
• deploying with google iLB, with dynamic support to add/remove nodes.
• deploying with elastifile LB, with dynamic support to add/remove nodes.
• Public IP, true/flase support.
• Custom configuration of the cluster.
• Full Destroy
• ILM configuration
• AsyncDR configuration
• The ability to run a deployment in 2 steps when needed, first creating the EMS (changing the deployment configurations...) and then resuming to full deployment.
• There are 3 option of LB
  • Elastifile LB, for this option use the following: - LB_Type: elastifile - LB_VIP: auto or a static IP address from outside the subnet.
  • Google iLB, for this option use the following: - LB_Type: google - We will automatically alocate a static IP within the subnet for the iLB
  • none, is used when you don't have enough credentials to configure the LB at the moment, and you will perform it after deployemnt. if you plan to configure the LB later on, use the following: - LB_VIP: auto or a static IP address from outside the subnet.
• There is an otption to use encrypted customer managed key for the EMS boot disk.

This version supports Elastifile Ver 2.7.x with the following:

• Dual replication for all configurations
• Adding nodes from a live cluster, by changing the SETUP_COMPLETE to true, and modifing the NUM_OF_VMS to the requested number. *** Removing nodes is not supported for version 2.7.5.x ***
• deploying with google iLB, with dynamic support to add/remove nodes.
• deploying with elastifile LB, with dynamic support to add/remove nodes.
• Public IP, true/flase support.
• Custom configuration of the cluster.
• Full Destroy

Small Local and Small Standard are not supported...