Salesforce Resolves 7.5-Hour Outage After Legacy Login Server Failure During Dreamforce
Updated
Updated · Computerworld · Sep 17
Salesforce Resolves 7.5-Hour Outage After Legacy Login Server Failure During Dreamforce
1 articles · Updated · Computerworld · Sep 17
Summary
Salesforce restored service at 2:59 p.m. EDT after a roughly 7.5-hour outage that began at 3:50 a.m., disrupting logins, triggering severe delays and intermittent errors, and blocking some new support cases.
An external dependency failure tied to a legacy login server drove up load on a core system component, stalling requests; rolling restarts and region-by-region fixes only partly worked before manual intervention restored remaining instances.
By 11:39 a.m. EDT, mitigations were in place across all instances, though some customers still reported missed scheduled jobs after access returned; a subset of Hyperforce instances were restored last.
Analysts said the incident exposed cloud resilience risks more than a simple legacy flaw, warning that hidden architectural dependencies can create platform-wide failures and leave data, workflows and integrations out of sync even after logins recover.
Salesforce apologized and said it will conduct a full investigation into the trigger, underlying cause and preventive actions, while experts said customers should now verify failed transactions, queued retries and downstream system consistency.
If modern cloud architecture promises high resilience, why did a single internal dependency cripple global operations during Salesforce's biggest event?
How many critical enterprise workflows remain broken and require manual reconciliation long after a cloud platform declares an outage resolved?
Could relying on automatic API versioning be the hidden trap that breaks your integrations when cloud platforms struggle to recover?