CRM Analytics went down for Salesforce customers for nearly 12 hours on 4 October, with Data Cloud ingestion and segmentation tasks dragged in later the same day. The fault sat inside Salesforce's own infrastructure, not in anything you control.
Started
13:07 UTC
Duration
Lasted 11h 54m
Source
IsDown
Next time
Salesforce is degraded. Is your product working?
An incident on their side does not always mean an outage on yours, and the only way to know is to already be watching your own endpoints. Set that up in under a minute and you will be notified the next time your own site is affected. And there will be a next time.
50 monitors for your own websites. No card.
Affected components: USA1, USA2S, USA3S, BRA6S, CAN26, NA223, CS252, BRA18, CS236, CS237, NA232, NA233, NA235, NA236, NA238, USA220, USA240S, USA242S, USA244S, USA248S, USA250S, USA252S, USA254S, USA256S, USA258S, USA260S, USA312, USA314, USA316, USA318, USA320, USA322, USA324, USA326, USA328, USA330, USA332, USA334, USA336, USA338, USA340, USA342, USA344, USA346, USA348, USA350, USA352, USA354, USA356, USA358, USA360, USA366S, USA372, USA374, USA376, USA384, USA80S, BRA34, BRA36, BRA44, BRA50, NA241, NA242, NA243, NA244, NA246, NA247, NA252, NA253, USA688, USA710S, USA716, USA722, USA732, USA744, USA746, USA748, USA754S, USA778, USA842, USA844, USA856, USA862, USA870S, USA872S, USA874S, USA876
What happened?
Salesforce logged the start of this incident at 11:15 UTC, with CRM Analytics jobs failing in the production environment and customers unable to use the feature at all. Across 87 components tracked that day, the disruption eventually touched 12, including USA1, USA2S, USA3S, BRA6S, CAN26, NA223, CS252, BRA18, CS236, CS237, NA232 and NA233. Salesforce's own engineers traced it to high memory utilisation in an underlying service, then to a supporting service that was slow to respond during startup, and spent the day adjusting resource limits and configuration rather than naming a single root cause. By 18:09 UTC Salesforce was warning that some customers could still see failures or delays running recipes, and that warning held through 20:06 and 22:23 UTC with no resolution. At 22:36 UTC the scope widened again: Data Cloud ingestion and segmentation tasks started failing or running late alongside the CRM Analytics problem. Just after midnight, Salesforce filed an impact radius increase, admitting the affected instance list was broader than first understood and that more customers would now get direct notifications. Recovery steps were reported complete at 00:34 UTC, with recipe runs starting to succeed again and work continuing on Data 360 segmentation tasks. The incident's logged duration runs to 11 hours 54 minutes from the 13:07 UTC start time used for tracking. No user reports were filed during this window on our side. Salesforce published 11 updates in total, which is a dense run for a single incident and reflects how long the underlying memory and startup problem resisted a clean fix.
Learning
This incident sat entirely inside Salesforce's backend, in the services that run CRM Analytics jobs and Data Cloud ingestion, a layer you cannot see or probe from outside. An uptime check against your own application endpoints would have stayed green throughout, because your endpoints were not what failed. The failure mode here is the one that catches people out most: a dependency buried inside a vendor's platform gets overloaded, recipe runs start failing, and nothing in your own stack shows a symptom until support tickets or failed jobs pile up. Salesforce's 90 day history shows 126 incidents with a median duration of 2 hours 32 minutes, so an incident running close to 12 hours with a scope increase mid-way is well outside the usual pattern for them. Our own 30 day figures show 31 incidents with a median of 2h 24m, which gives you a rough yardstick for how long a typical Salesforce disruption runs versus this one. Your own monitoring covers your half of the chain, the part you built. UptimeRobot now watches the provider's half too, so next time something like this happens you know which side broke without guessing.
