Salesforce Hit by Major Global Outage During Annual Dreamforce Event
San Francisco, Wednesday, 16 September 2026.
A major global outage disrupted Salesforce services on September 16, 2026, crippling enterprise operations worldwide precisely as the company hosted its flagship Dreamforce conference in San Francisco.
Technical Root Cause and Investigation
The service disruption was officially logged as Incident ID 20004433, beginning at 07:50 UTC on September 16, 2026 [4]. Salesforce engineers identified the root cause as an internal login service stalling requests, which subsequently consumed available server resources [4][5]. Initial investigations suspected an external dependency failure on the legacy login server, but third-party infrastructure providers confirmed no issues on their end [4]. This technical failure prevented end users from accessing affected services during the disruption window [1].
User Impact and Regional Spread
User reports indicated that the primary issue affected the Salesforce website, accounting for 69% of reports, followed by mobile apps at 16% and login services at 15% [1]. The outage impacted hundreds of instances throughout the world, including the USA, Japan, India, the UK, France, and Germany [5]. Volume of user reports surged around 9:05AM BST (4:05AM Eastern Time) according to outage tracking services [3]. This disruption coincided with the annual Dreamforce conference, where 40,000 people were expected to attend in person, with in excess of 200,000 people registered to attend online 240000 [5].
Resolution Efforts and Status Updates
Mitigation steps included blocking the API endpoint and attempting rolling restarts, though the latter was abandoned by 8:01 pm AEST on September 16, 2026 [4]. A fix was validated on a test instance as of 8:56 pm AEST on September 16, 2026, with a fleetwide rollout to all impacted instances currently underway [4]. IncidentHub data recorded 84 outages across 5,195 components within the 30-day period leading up to September 16, 2026 [2]. Salesforce committed to providing status updates every 30 minutes or sooner as investigations into the cause and restoration timeline continue [4][5].
Sources
- sundayguardianlive.com
- incidenthub.cloud
- community.designtaxi.com
- www.miragenews.com
- www.theregister.com