Cloud image process

Background

Current state of developing and maintaining GitLab images on the various cloud services is very distributed and not defined. This results in a slow pace of progress to creating images, lack of clear ownership regarding the development and maintenance of the images, lack of proper documentations on the images and several other challenges. I've discussed the topic to learn from how we worked on our AWS and Azure images to create a process that includes the lessons learned from those efforts.

Initial Action Items for Process Setup:

• Create GitLab Images vault in 1Password, invite the omnibus, partner marketing teams
• Store existing credentials for AWS and Azure
• Omnibus team to review existing images on AWS and Azure - evaluate if update is required
• Create email account used for this purpose - cloud.images@gitlab.com? subscribe omnibus, marketing and strategic partnerships to this email. Store in vault.
• Store marketing assets in shared folder:
  • short description of our service image
  • long description of our service image
  • Logos:
    • small square, 40 px
    • medium square, 90 px
    • large square, 115 px

Azure Image - Problems & Lessons Learned:

• First step of creating image is creating account: company information, logo, default settings of the VM on the cloud service
• Need to decide on pricing structure (byol model) - community and enterprise
• Store account details for each cloud provider in a shared location (AWS aren't stored anywhere shared)
• Push to staging, wait for input from cloud service - takes time to hear back
• Chase cloud support (option to escalate) - support is slow on the partner side
• Write messaging of the image on the cloud service - someone needs to own this task

I would love to get your feedback @sytses @ernstvn @JobV @marin @AshleyS to improve on this process. Remember that this is just the first version and one we will continually improve.