When a deployment fails —for example, a server crashes or network connectivity lapses—you can recover from the failed deployment by resuming or rolling back deployment:
Resume: The deployment operation picks up where it stopped and continues to completion. You can resume full and incremental deployments, and you can resume a deployment multiple times, in the event of successive failures.
Roll back: Target site deployment agents undo all work completed thus far in the failed deployment and restores the previous set of assets as the active set on the target. Only incremental deployments are candidates for roll back; a full deployment starts by deleting all target site data, which cannot be restored.
You can perform a rollback only once. If a failure occurs while rolling back a target site, the target site is then in an indeterminate state. In this situation you must perform a full deployment of the previous set of assets (created by one or more processes) to reestablish it as the active set on the target site.
After deployment recovery is complete, the target site is in a known, stable state and you can resume deploying to it as usual.
For information on full and incremental deployments, see Deployment Scope earlier in this chapter. For information on using the Admin Console in the ATG Business Control Center to stop, resume, and roll back deployments, and also for information on troubleshooting deployment errors, see Deployment Management.
For information on preventing and identifying errors in versioned data, see PublishingDataValidator.
