Re:Ask HN: What would your dream CI/CD pipeline tool be like?
by karmakaze on Sunday October 06, 2019 @03:49PM
(#21174601)
Push to a git repo which contains a dotfile indicating that CI/CD should act on it. Possibly from subdirectories if it has multiple artifacts or is in a monorepo.
Understands how to fetch/publish packages for common ecosystems (e.g. gems, jars, npm, go) and extensible to add others.
Define multiple environments with selectable automatic/manual progression to next environment upon passing test suite. Secrets can come from a separate repo with more restricted access. Each test/deploy should include the code and secret repo verion hashes.
Fast, parallel execution of tests in suite. Automatic retry of failed test, metrics collected to identify flakey code/tests.
Perform blue/gree deploys in each environment. Audible text-to-speech notification of test failures delivered to individual dev or group (via Slack and/or other means).
Quick one-button revert to an earlier version in any environment. Historical record of which version is running in every environment. Also have a button to force-promote a failed stage onto the next.
Periodic health checks running a subset of test suite with dashboard, alerts, and history. Audible text-to-speech announcement of health-check failures to dev/group.
It should deploy and test artifacts in the cloud of my choosing, in the deployment unit of my choosing (EC2, EKS, etc) to my own AWS (or GCP) account.
Understands how to fetch/publish packages for common ecosystems (e.g. gems, jars, npm, go) and extensible to add others.
Define multiple environments with selectable automatic/manual progression to next environment upon passing test suite. Secrets can come from a separate repo with more restricted access. Each test/deploy should include the code and secret repo verion hashes.
Fast, parallel execution of tests in suite. Automatic retry of failed test, metrics collected to identify flakey code/tests.
Perform blue/gree deploys in each environment. Audible text-to-speech notification of test failures delivered to individual dev or group (via Slack and/or other means).
Quick one-button revert to an earlier version in any environment. Historical record of which version is running in every environment. Also have a button to force-promote a failed stage onto the next.
Periodic health checks running a subset of test suite with dashboard, alerts, and history. Audible text-to-speech announcement of health-check failures to dev/group.
It should deploy and test artifacts in the cloud of my choosing, in the deployment unit of my choosing (EC2, EKS, etc) to my own AWS (or GCP) account.