Service Introspection Epic

[andrebriggs]

User Stories

As a developer/operator, I'd like to:

• [x] Spike to explore using service hooks to have AzDO notify us when pipelines are triggered and by whom
  • [ ] Push a message to azure storage queue when a build starts
  • [ ] Run a script that dequeues this message and gets the build id
    • [ ] Query the azure API to get variables for this specific build, to get image tag, hld commit id, etc.
• [x] Migrate CLI from Spektate to spk repository
  • [x] Move the models from spektate to its own npm package
  • [x] Make sure code follows strict rules
  • [x] Publish npm package so that the dashboard and CLI can use a common npm package 📦
• [x] Onboard the deployment introspection for my service(s) through CLI automation.
  • [x] CLI command onboard that will:
    • [x] Create a storage account (temporarily, until infra env template creates it) and add access keys in Azure Key Vault.
• [x] Enable observability by updating service deployment pipelines through CLI automation.
  • [x] Update the first pipeline (Source repo to Container registry) in the GitOps process to call service and update the table storage.
  • [x] Update the second pipeline (Container registry to HLD repo) in the GitOps process
  • [x] Update the third pipeline (HLD repo to Manifests repo) in the GitOps process
  • [x] Create a Variable Group and link to Azure Storage Keys
• [x] initialize the tool on a new machine for the first time with configuration values
  • [x] Explore all the different methods of storing private access keys as config, based on the global init command
• [x] spk deployment init using keyvault
  • [x] Support the option of supporting this initialization via Azure Keyvault
    • [x] Make sure keyvault is being used by pipelines via variable groups
    • [x] Make sure keyvault can be initialized via the init command
    • [x] Make sure non-keyvault initialization is still supported
  • [x] get a list of deployments by service
  • [x] Given the service name display the corresponding list of deployments
  • [x] Given a build id or commit id from source repository, display the deployments
  • [x] Given a release environment (such as Dev/Staging/Prod etc), display all the deployments
  • [x] Given an image tag, display all the deployments
  • [x] Given a deployment id, display the deployment
  • [x] Provide the option to output deployments in JSON, narrow or wide
• [ ] Enhance list of deployments
  • [ ] Given an author (into src repo), display all the deployments created by them
  • [x] Provide the option to --watch the deployments for a live view
• [ ] get a list of all the running logs in my deployment by service
  • [ ] a command that returns a snapshot of logs from a currently service deployment or optionally monitor/observe continuously as the logs get forwarded by passing a flag
• [ ] run a command to get additional information about a deployment which is not displayed in the CLI output table
  • [ ] A command such as spk deployment describe <deployment-id> that can act like a kubectl describe command which gives you detailed information about the deployment.
• [ ] run a command and get information about the deployment which is currently running on the cluster
  • [ ] Implement a command such as spk deployment cluster which displays the entire deployment in detail which is synced
• [x] run a program or a query based on the json output of service introspection CLI commands.
  • [x] create and print a json format of the output
  • [x] pretty print the output
• [x] run a command that allows me to validate my setup and onboarding of spk
  • [x] Make sure variables exist in AzDO
  • [x] Make sure Azure storage exists and the access keys to it are valid
  • [x] Make sure all configuration pre-reqs are present
• [x] Delete deployments from storage when their builds/releases have expired
• [x] Integrate the dashboard with the cli
  • [x] Host the dashboard in a docker container
  • [x] Perform a spike to figure out if the dashboard can be hosted in a web view within Azure DevOps
  • [x] Build a command in the spk tool to launch the dashboard, utilizing the global init

Related to #248

[andrebriggs]

One compromise for A way to "initialize" the service introspection tool is to create a shell script that:

• Creates an Az KV
• Creates Az Storage
• Puts Az Storage Key in Az KV
• Creates Variable Group in EXISTING AzDO project
• Links Az Stroage Keys to Variable Group

This script can then be easily used imported in a Terraform template and be instantiated here at some point in the future

[sarath-p]

@andrebriggs - we can make assumptions that the dependencies are in place for service introspection. For ex: cluster is already running with flux and AKV. We can create shell script or some other way for all new resources required/introduced for service introspection are created until they are implemented in right modules at some point.