Release management without a sandbox horror story

Four types, four refresh intervals, four data stories. Picking wrong costs you either fidelity or a 29-day wait.

1. VideoDeveloper, Developer Pro, Partial Copy and FullPreview6 min
2. ReadingRefresh intervals, and planning around a 29-day lock7 min
3. VideoSandbox data: seeding, masking, and the email that went to a real customer6 min
4. VideoScratch orgs, briefly, and when they beat a sandbox7 min
5. VideoChoosing a sandbox type by what you actually need to test6 min
6. LabLab: plan a release calendar around four sandboxes’ refresh windows6 min

Change sets, unmanaged packages and the metadata API. Plus the components that silently do not travel.

1. VideoChange sets end to end, including the validation-only deploy7 min
2. VideoInbound versus outbound change sets, and the org pair set up first6 min
3. ReadingWhat a change set cannot carry, and the workaround for each7 min
4. VideoReading a deployment failure honestly6 min
5. VideoPartial deployment, and the setting that is safer left off7 min
6. LabLab: deploy a flow, a field and a permission set, then reverse it6 min

Named credential passwords, remote site secrets, some picklist value sets and record ownership. Each is a post-deploy step in your plan.

1. VideoThe four things that never deployNamed credential passwords, remote site secrets, some picklist value sets and record ownership. Each one is a post-deploy step in your plan.6 min
2. VideoThe named credential whose password is blanked on every redeployRedeploying a named credential from source control blanks its password and silently breaks whatever depends on it.7 min
3. ReadingRemote site settings, and the callout that worked in the sandbox and not in production6 min
4. VideoRecord ownership and picklist value sets: the other two that do not travel7 min
5. LabLab: write the post-deploy checklist for a release with two of the four6 min

A concurrent deploy silently stripped a tab from the live app six minutes after ours shipped. Coordination is a release control, not a courtesy.

1. VideoThe deploy that stripped a tab six minutes laterA concurrent deploy silently stripped a tab from the live app six minutes after ours shipped, with no error on either side.6 min
2. VideoTwo deploys racing each other, and why the second one wins6 min
3. ReadingA local copy of an app file that was missing another team’s just-shipped tab7 min
4. VideoCoordinating a release window when more than one team can deploy6 min
5. VideoVerifying an app’s tab set after any deploy that touches it7 min
6. LabLab: reconstruct an app definition from two conflicting local copies6 min

Git without the software-engineering course attached. What it buys you, and the smallest useful workflow.

1. VideoWhat version control buys an admin who writes no code7 min
2. VideoRetrieving metadata, committing it, and reading a diff6 min
3. ReadingA deployment plan with a rollback step6 min
4. VideoBranches, and the smallest useful workflow for one admin7 min
5. VideoPackage.xml, and retrieving exactly what you mean to6 min
6. LabLab: put an org’s flows under version control7 min

A rollback plan someone else could execute, and a release calendar that survives a change freeze without blocking a genuine emergency.

1. VideoRunning a release under a change freeze without blocking urgent fixes7 min
2. VideoA rollback plan someone else could execute without you in the room6 min
3. ReadingThe release note that saves the next debugging session an hour7 min
4. LabLab: write a deployment plan with a rollback step for a real changeThe capstone lab. Plan a real deployment including the named credential and the tab it could silently affect.24 min
5. QuizCheck: release management10 min