Azure Key Vault secret reloads now tolerate individual failures

Azure Key Vault integrations in n8n had a reliability problem. When reloading secrets, disabled secrets were still being fetched — triggering unnecessary errors. More critically, a single inaccessible secret would abort the entire reload, leaving users with no secrets at all instead of the ones that worked.

Secrets marked as disabled are now skipped during the reload cycle, eliminating wasted API calls. More importantly, the fetch strategy changed from Promise.all to Promise.allSettled. Now when one secret fails, others still load. The system logs warnings for each failure without stopping. Only when every single secret fails does the reload throw an error — and even then, the previous cache remains intact.

These changes live in the external secrets module, specifically the Azure Key Vault provider. For users with large vaults containing a mix of active and archived secrets, or vaults where a handful of secrets occasionally become inaccessible, secret management becomes far more predictable.