Docs: Reflect resilient deployments on Octopus Cloud

[lucyjspence]

What

Deployments and runbook runs on Octopus Cloud now survive node restarts and maintenance windows — they pause and resume automatically rather than failing. Several docs still framed task failure during outages as expected behaviour, or described resilience as future roadmap work. This updates them.

Scope

Octopus Cloud only, deployments and runbook runs only. The claim is resilience shipped on Cloud, not maintenance windows removed — maintenance windows still exist; deployments and runbook runs now survive them.

Other task types (e.g. system tasks, health checks) are not resilient yet, so the maintenance-window page keeps the "other in-progress system tasks may still be abandoned" caveat.

Self-hosted HA pages are intentionally left unchanged — resilience is not yet on self-hosted.

Changes

File	Change
octopus-cloud/maintenance-window.md	Replace the "resilience is future roadmap work" hint with shipped behaviour; scope the shutdown-timeout note to deployments/runbook runs (other in-progress system tasks may still be abandoned)
octopus-cloud/frequently-asked-questions.md	Add "What happens to my deployments during a maintenance window?" under Availability
octopus-rest-api/.../rerun-deployments-and-runbooks-after-node-shutdown.mdx	Add a Cloud-obsolescence note; script remains relevant for self-hosted HA

🤖 Generated with Claude Code

[xwipeoutx]

Added suggestions for "continue" vs "pause and resume", to avoid confusion with manual intervention and the like

[xwipeoutx]

Suggested change

	Deployments and runbook runs on Octopus Cloud are resilient. If a [maintenance window](/docs/octopus-cloud/maintenance-window) or node restart interrupts a deployment or runbook run, it continues automatically once your instance is back online, rather than failing. You don't need to drain tasks or manually re-run them afterwards.

[xwipeoutx]

Suggested change

	Upgrading an instance is the primary cause of outages. Deployments and runbook runs on Octopus Cloud are now resilient: if an outage interrupts a deployment or runbook run, it continues automatically once your instance is back online, rather than failing. You no longer need to drain tasks or manually re-run deployments and runbook runs after a maintenance window.

[xwipeoutx]

Suggested change

	- Your instance will be given a few minutes to shut down cleanly. This will allow any in-progress tasks to complete. Deployments and runbook runs still in progress at the end of the timeout are resilient — they continue automatically once your instance is back online, so they no longer fail. Other in-progress system tasks may still be abandoned.

[xwipeoutx]

Suggested change

	**Octopus Cloud:** This script is no longer needed for deployments and runbook runs. On Octopus Cloud, these tasks are resilient — they continue automatically after a node restart or maintenance window rather than being cancelled. This script remains useful for self-hosted High Availability instances, where node restarts still cancel active tasks.

[team-marketing-branch-protections]

Pull request environment is available at https://stoctodocspr3218.z22.web.core.windows.net.

You can view the ephemeral environment status in Octopus Deploy.

This environment will be automatically deprovisioned when the pull request is closed, or after 7 days of inactivity.

[lucyjspence]

Incorporated @steve.leigh's suggestions — "paused/resumes" → "continues" in all four places (maintenance-window hint + outage bullet, Cloud FAQ, rerun-script note). Thanks for the precise-language nudge.

Two intentional differences from the literal suggestions:

• Kept the "other task types are not yet resilient" caveat. >99% of task executions on Cloud are on V2, but that's by volume, not task-type coverage — non-deployment/runbook tasks (health checks, system tasks) aren't resilient yet. Without the caveat the page reads as "everything survives an outage," which would overclaim.
• Removed the "drain tasks" wording. Octopus drains the instance during maintenance on the customer's behalf — they never did it themselves, so "you no longer need to drain tasks" implied an action that was never theirs. Kept "no longer need to manually re-run them," which is a real customer action.

Also fixed the failing lint check — pre-existing MD007 (4-space nested bullets) and MD047 (missing trailing newline) in the rerun-script page, surfaced because this PR touches the file. Lint + spellcheck green.