Pause actions
pause_workflow and pause_credential always send a run.stopped notification through your notification rules. Rules that listen for run.failed receive it too, so existing rules keep working. It names the error code, what was paused, and how many workflows and queued runs were stopped or held, so a pause is never silent. A workflow with notifications turned off stays quiet.
Pause the credential on login failure
Three login failures can pause the vault credential the run used, so queued runs stop retrying a login that cannot work and the account is not locked out. It is off by default and you opt in under Settings → Error Handling → Pause credential on login failure:
Turn on any of them. The run that hit the failure ends as
execution.stopped. If the run used several vault credentials, only the one for the site it failed on is paused; when that site matches none of them, or more than one, nothing is paused. A workflow can use the workspace setting or override it on its own Error Handling tab, or through the Workflow API with pause_on_invalid_credentials and pause_on_password_update. Omit a field to leave it unchanged, send null to use the workspace setting, and send false to turn it off for that workflow even when the workspace has it on. The workspace-wide setting can only be changed in the dashboard. A skippable password prompt never pauses.
Password update required cannot be on while Password Update Recovery is on, because recovery changes the password itself. Turn one off to use the other. Unpause the credential in Runs → Paused Resources, or call POST /runs/unpause with pause_resource set to credential.
An error code with its own action, such as pause_workflow or pause_credential, takes priority over this setting.
Pause the workflow when the service is retired
When a run fails because the site says the application or service itself has been permanently retired, discontinued or decommissioned (SERVICE_UNAVAILABLE / SERVICE_RETIRED), the workflow can be paused so queued runs stop failing against a service that no longer exists. It is off by default and you opt in under Settings → Error Handling → Pause Workflow on Retired Service. A workflow can use the workspace setting or override it on its own Error Handling tab.
The run that hit the retirement notice ends as execution.stopped, queued runs of that workflow stay in execution.paused, and the pause shows in Runs → Paused Resources with the reason “Paused because the target service is retired”. It never pauses a credential, because a retired service says nothing about the account. Scheduled maintenance, a temporary outage or a single unavailable item is not a retirement and never pauses. Unpause the workflow in Runs → Paused Resources, or call POST /runs/unpause with pause_resource set to workflow.
Every pause from a pause setting sends a run.stopped notification that says why it paused, for example “Paused 1 workflow because the target service is retired”. An error code with its own action takes priority over this setting.