- 7 years of devop
- docker
- kubernetes
- Help to move to more automated processes
- Worked in azure devops support
- Working on pipeline issues
- AWS
- IAM
- VCP
- EC2
- S3
- Cloudwatch
- Why interested?
- Helping customer and platform working learning on how to use it.
- Previous project ended
- Question 8:
- How to package the messages
- What things should be included and excluded
- Let the customer know about the impacts
- Let to know what will be applied
- Avoid unconfirmed cause
- Avoid technical information
- Questions 4
- Would find out
- What the automate steps are
- What the manual steps
- Look into most customer critical place to verify
- Would
- Start to standardize the process
- Check on environment drift
- Add pre and post deployment checks
- Try to make a more control automation
- How would you handle customer specific configurations
- Try to externalize the customer configuraions
- Use a customer specific variable group
- Would have a customer specific set of configurations with secret managers
- Approach
- For an orchestrated deployment
- Statge 1: Use cloudformation
- Stage 2: Installation
- Stage 3: Customer specific changes
- Keep them externalized
- Stage 4: Add in auditability
- Could add into an orchestration process
- Have a parent pipeline with different modules
- Question 3
- Check what changes have been made lately
- Check on timeouts
- Check on logs
- See if there were changes in endpoint urls
- Check with dns resolution
- Check on the host name
- Check that the target point resolves
- Check on the Ec2 instance
- See if the role has changed
- Check on the policy
- Check on the loader balancer
- Check on the configuration
- Check on the path
- S3 bucket connection
- Start on the ec2 instance
- try aws sts
- Check if the role has changed
- Check the ec2 instance role.
- Can the role see the instance
- NOT answering how to check if the role has been changed
- Can check on cloudtrail for changes
- Having a single processes makes it hard to control faiures
- Question 1
- Check on the recent change history
- Check on app dependendies, permission
- Look at the exit
- Look at docker ps to check on the docker logs
- See if there is a permission error
- Compare between working and non-working version to see what differences they are
- Check on the processes in the container
- Would use docker exec
- Start with docker logs
- Check on docker ps
- Look in docker exec
- Can check the environment variable
- Can check whether the dirctories are mapped correct
- Can check the file permissions
- Can check whether it can connect to the outside resources it needs
- Check the mounted paths
- Can check the container config
- How to compare two dockers
- Check the image tags
- Check the start commands
- Check the exposed ports
- Check the outputs for the docker that is failing against the working one.
- Can check with docker diff to see what changes
- Can recreate the image from a known working one
- Cna focus on the image and variable and dependencies
Question 2
- Check if there is an outage
- Test the file from end to end
- Find where the file stops, transform wize
- Try to compare the file that fails with successful files
- Check if there a delimiter issues
- Check how the workers are working on that file
- How to prepare evidence for root cause analysis
- Try to avoid changing the original file and making a working copy
- Have some rows before and after
- Try to find the issue on the row number
- Will try to make the smallest changes possible
- Check on the row count
- Try to reprocess from the erred steps