Prerequisites for Kubernetes

• Kubernetes cluster setup: Creating an Amazon EKS cluster
• DockerHub repository must be reachable from nodes.
• SSL certificates: Valid SSL certificates as per the DNS of the Core server, DB server, and the Kubernetes cluster.
• Kubectl, helm3 packages installed on the core server for installing pods and necessary permission should be available for deploying containers on the Kubernetes cluster.
• Helm charts downloaded to the core server into /tmp/{jiffy-install}/{helm}

  • wget –user {username} –ask-password downloads.jiffy.ai/4.5/Release/jiffy-helms-aws-V4.5.zip
  • unzip jiffy-helms-aws-V4.5.zip

Prerequisites for Core Server and DB Server

• SSH connectivity between core server and DB server (preferred method using the private key).
• EFS volumes to be mounted into the following paths. *{core server mountpoint}/jiffy/auth/ *{core server mountpoint}/jiffy/ml/config/
• Connected users must have sudo privilege (For root installation).

• Repositories/Internet access should be provided on the DB and Core machines. *Recommended repositories:

  For RHEL Users	For CentOS
  rhel-7-server-rpms	epel-release
  rhel-7-server-extras-rpms	centos-release-scl
  rhel-7-server-optional-rpms	docker-ce
  rhel-server-rhscl-7-rpms	pypi.org
  docker-ce
  docker-ce
  pypi.org

• Extract the ansible-playbook tar file in the core server (tar -xf jiffy-playbook.tar.gz), as normal user. The extracted files include the inventory and variable.yml.

• Inventory file filled as per the user environment(contains all the details of the machines, authentication method).

• Variables file filled as per the user environment(contains all the parameters, such as, username, configurable values, mountpoint, etc.).

• Execute the following command to download Artifacts to a new folder under /tmp/{jiffy-install} in core machine.

wget --user {username} --ask-{password} downloads.jiffy.ai/4.5/Release/jiffy-installation.tar.gz

sha1sum: 00368671024b5f3cbe367e53ac9c7efcc0676fd2

md5sum: da8eedc7ad19f910ff39adb49168741b

wget --user {username} --ask-{password} downloads.jiffy.ai/4.5/Release/jiffy-playbook.tar.gz

sha1sum: 8c96cfbd4aaa319818ac1abc465df7038986d4bd

md5sum: f0c5203c797859eba1ae2e705584a95e

Installation Steps

Kubernetes deployment

1. Update efs mount point values in pvcreation.yml prior to triggering the deployment.
   1. volumeHandle: {EFS ID}:/jiffyauthdata
   2. volumeHandle: {EFS ID}:/mldata/config
2. Run the following commands to create persistent volume, persistent volume claim, and deploy Kubernetes pods.

Create namespace jiffy-cognitive

kubectl create namespace jiffy-cognitive

• kubectl apply -f pvcreation.yml -n jiffy-cognitive
• kubectl apply -f pvc-creation.yml -n jiffy-cognitive
• kubectl apply -f imagepull-creds.yml -n jiffy-cognitive

• helm install portfolio portfolio/ –set image.tag=JARVIS-MAIN-Portfolio_18807Sep21 –set image.repository=jiffyai/portfolio –set replicaCount=1 -n jiffy-cognitive
• helm install pdfsplit pdfsplit/ –set image.tag=JARVIS-MAIN-Pdfsplit_2007Sep21 –set image.repository=jiffyai/pdfsplit –set replicaCount=1 -n jiffy-cognitive
• helm install bolml bolml/ –set image.tag=JARVIS-MAIN-T4.3.011807Sep21 –set image.repository=jiffyai/bolml –set replicaCount=1 -n jiffy-cognitive
• helm install bol-category bol_category/ –set image.tag=JARVIS-MAIN-vendorcategorisation_6407Sep21 –set image.repository=jiffyai/bolcategory –set replicaCount=1 -n jiffy-cognitive
• helm install handwritingsegmentation handwritingsegmentation/ –set image.tag=JARVIS-MAIN-HWSegmentaion_3802Sep21 –set image.repository=jiffyai/hwsegmentation –set replicaCount=1 -n jiffy-cognitive
• helm install invoicecategory invoicecategory/ –set image.tag=JARVIS-MAIN-T4.3.011807Sep21 –set image.repository=jiffyai/invoicecategory –set replicaCount=1 -n jiffy-cognitive
• helm install pdf2image pdf2image/ –set image.tag=JARVIS-MAIN-Pdf2Image_2107Sep21 –set image.repository=jiffyai/pdf2image –set replicaCount=1 -n jiffy-cognitive
• helm install pdf2json-service pdf2json-service/ –set image.tag=JARVIS-MAIN-Pdf2JsonService_1901Sep21 –set image.repository=jiffyai/pdf2json-service –set replicaCount=1 -n jiffy-cognitive
• helm install template-converter template-converter/ –set image.tag=JARVIS-MAIN-104 –set image.repository=jiffyai/template-converter –set replicaCount=1 -n jiffy-cognitive
• helm install w2split w2split/ –set image.tag=JARVIS-MAIN-W2Split_4807Sep21 –set image.repository=jiffyai/w2split –set replicaCount=1 -n jiffy-cognitive
• helm install backward-comp backward-comp/ –set image.tag=JARVIS-MAIN-BackwardCompatibilityEngine2307Sep21 –set image.repository=jiffyai/backward-compatibility-engine –set replicaCount=1 -n jiffy-cognitive
• helm install functionator functionator/ –set image.tag=JARVIS-MAIN-116 –set image.repository=jiffyai/functionator –set replicaCount=1 -n jiffy-cognitive
• helm install invoiceml invoiceml/ –set image.tag=JARVIS-MAIN-T4.3.011807Sep21 –set image.repository=jiffyai/invoiceml –set replicaCount=1 -n jiffy-cognitive

kubectl create secret tls tlssecret1 –key {key path of Kubernetes} –cert {cert path of Kubernetes} -n default

kubectl apply -f ingress-nginx.yaml

Core and DB Server Deployment

1. Change directory to the installation path, where the download artifacts are present.

   cd /tmp/{jiffy-install}

2. Execute the following command to install ansible and dependent packages in the core server:

   chmod u+x preAnsibleSetup.sh
   ./preAnsibleSetup.sh

3. Update the inventory file. Click here to download a sample Inventory file.
4. Configure the variable.yml file in the /tmp/{jiffyinstall}folder as per the corresponding environment. Click here to download a sample Variables file.
5. Execute the play book to begin the installation and follow instructions on the screen.

   ansible-playbook main.yml -i inventory -e @variable.yml

   For debug mode,

   ansible-playbook main.yml -i inventory -e @variable.yml -v

   For detailed debug mode,

   ansible-playbook main.yml -i inventory -e @variable.yml -vvvv

6. Update privileges for the vault bootstrap script.

   chown {jiffyuser}:{jiffyuser} vaultBootstrap.sh

7. Switch to Jiffy user to initialize the vault.

   • sudo su {jiffyuser}
   • chmod u+x vaultBootstrap.sh
   • ./vaultBootstrap.sh

8. Follow screen prompt to complete vault initialization.
9. Switch back to sudo user.
10. Update variable vault_token: {replaceme} in variables.yml with vault token.
11. Execute the below ansible command to bootstrap application:
    1. with masterkey prompt:

       ansible-playbook jiffybootstrap.yml -e @variable.yml

    2. without masterkey prompt:

       ansible-playbook jiffybootstrap.yml -e masterKey=‘{replaceme}’ -e @variable.yml

Generate Self Signed Certificates

(Use only in non-prod environments as a temporary measure if valid SSLs are not available.)

1. To Generate SSL Certificate for DB Instance and Cognitive Instance:

   1. Navigate to /tmp/{jiffyinstall}; folder.
   2. Execute the following commands:

      chmod +x generate-db-cert.sh
      ./generate-db-cert.sh -d {DBInstanceDomainname}

   3. The following files are generated:
      • db_ca_key.pem
      • db_ca_cert.pem
      • dbcn_cert.pem
      • dbcn_key.pem
        

2. Localhost certificates for the core server are generated by ansible and are available under /tmp/{jiffyinstall}

   • ca_key.pem
   • ca_cert.pem
   • localhost_key.pem
   • localhost_cert.pem