sunset

danger

DON'T PERFORM THIS ACTIONS UNLESS IT IS MANDATORY
Get the strong confirmation from the DevOps Lead

This doc explains about the sunset operation.
Sunset of a server is to plan to intentionally remove or discontinue it.
There are two types of sunset.
- Hard sunset : In this we take complete backup of all data and terminate all the resources. This is more cost effective than soft sunset but takes longer and more efforts to restore.
- Soft sunset : In this we just stop the instances instead of deleting/terminating them. We do delete any resources that have a fixed monthly charge..(like load balancers) and are easy to spin up again. This allows very fast and hassle-free/restoration, in case the setup needs to be made live again.

Hard Sunset

Cost Incurred: In this we incur charge only for the data backed up to S3. Below is the monthly cost for a typical brand(varies with the amount of data being backed up)
- single_stack (dev): ~$0.5/Month
- multi_stack (prod): ~$1-2/Month
Use Case: When a setup needs to paused/turned off for an unknown duration of time. Typically for more than a month.
- A setup was requested, but there's delay from client side.
- Termination of client contract.
- End of a POC.
- etc.
Restoration timelines: Typically, add a day to the below mentioned timelines as these timelines does not include the tasks the team might have a planned.
- single_stack (dev): ~1-2 days
- multi_stack (prod): ~2-3 days

Soft Sunset

Cost Incurred: In this we incur charge only for the EBS volumes of the servers. Below is the monthly cost for a typical brand.
- single_stack (dev): ~$3/Month
- multi_stack (prod): ~$6-9/Month
Use Case: When a setup needs to paused/turned off for know duration of time in the range of up to ~2-3 weeks.
- A setup was requested, but there's delay from client side.
- A client contract ended, but the data might be required for report generation or similar activity.
- etc.
Restoration timelines: Typically, add a day to the below mentioned timelines as these timelines does not include the tasks the team might have a planned.
- single_stack (dev): ~1-2 days
- multi_stack (prod): ~2-3 days

single_stack

common_steps

These are the common steps need to perform for both soft and hard sunset.
single stack contains the Fullstack and devAI servers.
The sunset process for single_stack refers to the
- Login to the Fullstack server of respective brand.
  - Take the backup of the existing data from the mongodB container
  - create a directory db-backup on home directory.
  - Copy the content of the database into the newly created folder for backup db-backup.
  - Now, copy the content of the db-backup folder to the s3://ori-db-backup/brand_name/Backup/database/Development/oriserveDemoDB/ (change the brand_name as per the requirement.)

You can use the below script for taking the backup of the database and copy to the S3 bucket.

#!/bin/bash
echo "Please enter DB service type : "
read -rp "Default[docker] : " dbServiceType
[[ -z ${dbServiceType}  ]] && dbServiceType="docker"
echo "Please enter DB name : "
read -rp "Default[oriserveDemoDB] : " DBName
[[ -z ${DBName}  ]] && DBName="oriserveDemoDB"
echo "Please enter DB user : "
read -rp "Default[dbOwner] : " DBuser
[[ -z ${DBuser} ]] && DBuser="dbOwner"
echo "Please enter DB user's password : "
read -rp "Default[oriSaaSdemo1092] : " DBuserpass
[[ -z ${DBuserpass} ]] && DBuserpass="oriSaaSdemo1092"
echo "Please enter S3BucketName to upload backed up DB : "
read -rp "Default[ori-db-backup] : " S3BucketName
[[ -z ${S3BucketName} ]] && S3BucketName="ori-db-backup"
echo "Please enter ProjectName : "
read -rp "Exits on empty string : " ProjectName
[[ -z ${ProjectName} ]] && echo "ProjectName is required. Exiting.." && exit 1
echo "Please enter Environment : "
read -rp  "Default[Development] : " Environment
[[ -z ${Environment} ]] && Environment="Development"
#echo "${Development} ${DBName} ${DBuser} ${DBuserpass} ${S3BucketName} "
############ BackupDB ############################################
sudo mkdir -p /backupDB
cd /backupDB
if [[ ${dbServiceType} = 'docker' ]]; then
   ############### DOCKER ###################
   #mongodump -d ${DBName} -h 127.0.0.1:27017 -u ${DBuser} -p ${DBuserpass} --authenticationDatabase ${DBName} --gzip -o ./
   docker exec mongoDB mongodump -d ${DBName} -h 127.0.0.1:27017 -u ${DBuser} -p ${DBuserpass} --authenticationDatabase ${DBName} --gzip -o ./
   [ "$?" -eq 0 ] && sudo docker cp mongoDB:/${DBName} /backupDB/${DBName}  || exit 1
   [ "$?" -eq 0 ] && (docker exec mongoDB [ -d "/${DBName}"  ]  && (echo "dump dir exists in container" && exit 0) || (echo "dump dir doesn't exist" && exit 1)) || exit 1
   [ "$?" -eq 0 ] && (docker exec mongoDB rm -rf /${DBName} && echo "dump dir removed from container" || echo "unable to remove dump dir from container")
   ############### DOCKER ###################
else
   ############### SERVICE ###################
   mongodump -d ${DBName} -h 127.0.0.1:27017 -u ${DBuser} -p ${DBuserpass} --authenticationDatabase ${DBName} --gzip -o ./
   ############### SERVICE ###################
fi
#################### UploadtoS3 #####################################
[ "$?" -eq 0 ] && echo uploading to s3
aws s3 cp /backupDB/${DBName} s3://${S3BucketName}/${ProjectName}/Backup/database/${Environment}/${DBName}/ --recursive
#aws s3 cp /backupDB/${DBName} s3://${S3BucketName}/${ProjectName}/Backup/database/${Environment}/${DBName}/ --recursive --storage-class GLACIER
echo removing local backup directory
sudo rm -rf /backupDB

Run the script with sudo privilages.
While running the script you need to provide some details like DBname, DB user, DB user's password, s3bucketName, projectname/brandName, Environment.

JENKINS_JOBS_BACKUP

Now, we need to take the backup of the respective brand jobs.
Login to the jenkins server.
- Go to the location /var/lib/jenkins/.
- check for the backup folder. If it is not there create a directory jobBackup.
- Create a new folder for the brand_backup env_brandName_jobs inside the jobBackup
- Switch to jenkins user sudo su jenkins
- Let's say brandName is tata_capital create a directory named dev_tata_capital_jobs
- Copy the all jobs of tata_capital to the tata_capital_jobs directory.
  - command : cp -a /var/lib/jenkins/jobs/dev_tata-capital_* /var/lib/jenkins/jobBackup/dev_tata_capital_jobs/
Now, you have taken backup of all jobs of required brand.
Zip the folder tata_capital_jobs
- command : zip -r dev_tata_capital_jobs.zip dev_tata_capital_jobs
You will find a folder in s3 under the respective brand backup
Upload this zip folder to the location s3://oriserve-demos/tata-capital/Backup/Environment/Development/Jenkins/
Replace the Environment according to the requirement.
Disable the respective jenkins jobs.

disable_jobs

PNG_DASHBOARD

After deleting all the resources,
Login to the grafana
Whenever you opened the Grafana, click on the General, top left.
Search the brand you want to delete.
Search for the respective dashboard, here incase air-arabia dashboard. You can see in below image reference.
Delete all the monitoring dahsboards. Below image is for only ref.
Click on the specification you want to delete, click on the Remove and confirm the removal.
Delete all the dashboards under repective brand.

soft

After successfully taking the backup,
Enable the TerminationProtection.
- Go to Actions --> Instance settings --> Change instance termination protection and select enable.

temination_protection

Stop the Fullstack, AI instances as below.
- Go to Instance state --> Stop Instance

bot_heart_beat

For bot_heart_beat process of soft sunset. Refer to the, bot_heart_beat

hard

If the stack is created through cloudformation , delete the stack through cloudformation only.Below image is for reference.
Go to the AWS console and search for the cloudformation service.
Search for the required brand. Let's say Baja-thai

stack_delete

If the stack is created manually, delete the instances manually.

S3 archival

Archive (move) the deleted brands to the Archieved/ folder in the oriserve-demos bucket.
- Before moving the brand delete all the codeDeploy deployments from all services folder.

bot_heart_beat

For bot_heart_beat process of soft sunset. Refer to the, bot_heart_beat

Access_management_changes

danger

Before performing this action, Make sure all the environmets has been deleted/stopped.
If you are not sure, get the confirmation from DevOpsLead

After performing the necessary hard sunset operation, we need to update the brand details to sunset folder,which indicates that particatular brand has deleted.
In your local, clone the access_management repo.
In the repo, we will find the projects folder where all our projects access details will present. Let's say our brand is baja-thai
- Location to our brand bajaj-thai will be projects/bajaj-thai
We will find the sunset folder along with the projects inside the access_management repo.
If we need to sunset the bajaj-thai brand . Move the Bajaj-thai brand from projects folder to sunset folder.
- Command : mv projects/bajaj-thai sunset/
After updating, commit the changes and raise a PR.
- commands : git add .
- git commit -m "your message"
- git push -u origin <your branch>
After pushing the changes, raise a PR.
DevOps Lead will review the changes and will merge.

multi_stack

common

These are common steps need to perform for both soft and hard sunset.
Multistack contains the Utility, NLP , Webapp, dB servers.
Hard sunset refers to the terminating the Utility, NLP , Webapp, dB servers after taking the backup of the database.
soft sunset refers to the stopping the Utility, NLP , Webapp, dB servers after taking the backup of the database
The sunset process for multistack refers to the
- Login to the DB server.
  - Take the backup of the existing mongodB container.
  - For sunset operation, we have a script.
  - create a directory db-backup on home directory.
  - For script, Please refer the above script. script is same for single stack and multistack.
  - Run the script with sudo privilage.
  - While running the script you need to provide some details like DBname, DB user, DB user's password, s3bucketName, projectname/brandName, Environment.

JENKINS_JOBS_BACKUP

After successfully taking the backup of the database, take the jenkins_job backup refer the jenkins_job_backup

PNG_DASHBOARD

After deleting all the resources,
Login to the grafana
Whenever you opened the Grafana, click on the General, top left.
Search the brand you want to delete.
Search for the respective dashboard, here incase air-arabia dashboard. You can see in below image reference.
Delete all the monitoring dahsboards. Below image is for only ref.
Click on the specification you want to delete, click on the Remove and confirm the removal.
Delete all the dashboards under repective brand.

soft_sunset

After successfully taking the backup,
- Enable the TerminationProtection.
- Go to Actions --> Instance settings --> Change instance termination protection and select enable.
- Stop the Utility, dB instances. webApp and AI servers will be done by ASG which is explained below.

instance_stop

Delete the Webapp, AI load balancers, by enabling the Deletion protection (if it is not enabled.)
- Load_balancers naming will be brandName-prod-ai-ALB , brandName-prod-webApp-ALB
- Here, bajaj-thai-prod-ai-ALB, brandName-prod-webApp-ALB
For Webapp and NLP, update the autoscaling configuration with Desired,minimum and maximum capacity as zero. So that, it will automatically delete the respective instances.
- AutoscalingGroups naming will be brandName-prod-AI-ASG, brandName-prod-webApp-ASG
- Here, bajaj-thai-prod-AI-ASG, bajaj-thai-prod-webApp-ASG

ASG

bot_heart_beat

In soft sunset, we need to redirect the alarms to dumped-grafana-alerts channel in slack.
Login to the grafana
You can see the folder chatbot as below
Get into the chatbot folder, you will find the heartbeats like below
Scroll down , you can see the alerts like below
Click on the Edit. It will redirect to the page like this
In the Alert section, for Sendto option,
- Remove the existing alerts.
- Add the dumped-grafana-alerts channel and remove all other channels.
Save the Alert .

hard_sunset

If the stack is created through cloudformation , delete the stack through cloudformation only.Below image is only for reference.

stack_delete

If the stack is created manually, delete the instances manually.
Delete the load balancer,by disabling the deletion_protection.
- Select the load balancer.
- on the Description tab, choose Edit attributes
- On the Edit load balancer attributes page, select Disable for Deletion Protection, and then choose Save.

Launch_Template_Backup

We use LaunchTemplates for 2 services.
- Webapp and NLP
We need to take the backup of the launch template data from NLP and WebApp LaunchTemplates.
- command : aws ec2 describe-launch-template-versions --versions '$Latest' --launch-template-id <--launch template-Id--> --query "LaunchTemplateVersions[*].LaunchTemplateData" > LT.json
It will save the launch template data of the latest version into the LT.json file.
Upload this LaunchTemplate backup file into S3. Location will be as follows.
- For NLP : s3://oriserve-demos/brandName/Backup/Environment/Production/LT_Data/oriNlp/
  - Command : aws s3 cp LT.json s3://oriserve-demos/brandName/Backup/Environment/Production/LT_Data/oriNlp/
- For Webapp : s3://oriserve-demos/brandName/Backup/Environment/Production/LT_Data/webApp/
  - Command : aws s3 cp LT.json s3://oriserve-demos/brandName/Backup/Environment/Production/LT_Data/webApp/

bot_heart_beat

In hard sunset, we need to delete the bot_heartbeat alerts.
Login to the grafana
You can see the folder chatbot as below
Get into the chatbot folder, you will find the dashboards like this
Scroll down to the page, you can we see the configured alarms, as shown below.
You will get the options like above.
Click on the Remove and delete the alarms.

danger

Hard Sunset

Soft Sunset

single_stack​

common_steps​

JENKINS_JOBS_BACKUP​

PNG_DASHBOARD​

soft​

bot_heart_beat​

hard​

S3 archival​

bot_heart_beat​

Access_management_changes​

danger

multi_stack​

common​

JENKINS_JOBS_BACKUP​

PNG_DASHBOARD​

soft_sunset​

bot_heart_beat​

hard_sunset​

Launch_Template_Backup​

bot_heart_beat​

single_stack

common_steps

JENKINS_JOBS_BACKUP

PNG_DASHBOARD

soft

bot_heart_beat

hard

S3 archival

bot_heart_beat

Access_management_changes

multi_stack

common

JENKINS_JOBS_BACKUP

PNG_DASHBOARD

soft_sunset

bot_heart_beat

hard_sunset

Launch_Template_Backup

bot_heart_beat