Salesforce Outage on 16 September 2026: Global Disruption Hits During Dreamforce

What happened in the Salesforce outage

A global Salesforce outage struck on 16 September 2026, starting at 07:50 UTC — 1:20 pm in India — and hitting multiple instances in every region, according to status updates republished by Mirage News. The timing was awkward: the failure landed in the middle of Dreamforce, Salesforce's flagship three-day conference at San Francisco's Moscone Center. In India, searches for "salesforce" jumped sharply, with Google Trends logging roughly 2,000 queries over 24 hours — a measure of curiosity, not evidence about the cause.

Users reported long delays, sporadic errors and stretches where services simply would not load. Adding to the frustration, the Help portal stopped accepting new support cases, so administrators could not even file tickets about the problem they were experiencing.

Why the login service failure cascaded

Salesforce's own investigation pointed at an internal component rather than an outside provider. Engineers found that incoming requests were hanging while they waited for a response from an internal login service, and the backlog steadily consumed available server capacity. Think of it as a single jammed toll booth backing up an entire highway: every new request queued behind the stalled ones until the servers had nothing left to give.

An early rolling restart of one affected instance did not help, and the team abandoned restarts as a remediation strategy. Salesforce also said it checked with its third-party infrastructure provider, which reported no problems on its side — effectively ruling out an upstream cloud failure. The incident was tracked under ID 20004433.

Timeline of the response (16 September 2026)

• 07:50 UTC (5:50 pm AEST): Disruption begins; end users lose access.
• 7:10 pm AEST: Internal login service identified as the bottleneck; restart attempted.
• 8:18 pm AEST: A core component under heavy load is cited as limiting request processing.
• 8:56 pm AEST: A fix validated on a test instance; fleetwide deployment begins.
• 9:41 pm AEST (11:41 UTC): Rollout proceeding region by region, with service improving but no confirmed completion time.

At the last update available when this was written, Salesforce said customers were beginning to see normal service return while engineers worked on a permanent code-level fix. The company had not declared the incident closed, so the definitive word should come from trust.salesforce.com, including any post-incident summary.

GovCloud spared; Dreamforce continued

Salesforce confirmed that GovCloud — its environment for US government workloads — stayed outside the impact zone. Meanwhile Dreamforce carried on, where the company had just unveiled AIforce, which lets AI systems operate on Salesforce data, workflows and business logic, plus Claudeforce, a beta integration with Anthropic's Claude assistant.

What Indian Salesforce teams should take away

India's large base of Salesforce developers, administrators and offshore delivery teams felt this outage in the middle of the working afternoon. Practical lessons:

1. Bookmark trust.salesforce.com and check your own instance's status rather than relying on social media chatter.
2. Reference incident ID 20004433 when filing cases once the Help portal is reachable again.
3. Review integration retry and timeout logic — this failure mode was stalled requests, not clean error responses, which can hang poorly written connectors.
4. Wait for Salesforce's formal post-incident report before drawing conclusions, since no definitive root cause had been published at the time of writing.

Separately, same-day reports mentioned possible Google service issues, but Google Cloud's own dashboard showed no broad severe incidents as of 12:01 PDT on 16 September, so no connection between the two is confirmed.