Scheduled maintenance: cache tier in US East and US South Central
resolved- Started
- Oct 4, 7:00 AM EDT
- Resolved
- Oct 4, 9:00 AM EDT
- Duration
- 2h
Cerebrium's updates (2)
- update · Oct 4, 7:00 AM EDT
On Sunday, 4 October 2026, at 07:00 EDT, we'll be carrying out scheduled maintenance on the distributed cache tier in US East and US South Central. This cache speeds up reads of the model weights and files your apps load when a replica starts. The cache will remain in service for the duration of the window and no stored data is changing or at risk. While the cache rebuilds, during the window and for several hours after it, the first read of a given file in these clusters may be slower than usual. Apps that load model weights or other files at replica start may see longer cold starts, mostly at the slow end rather than the typical start. Repeat reads return to normal speed. Running replicas, live request handling, builds, container image pulls, the dashboard, the REST API, metrics, us-central1 and all EU regions are not affected. If your app is latency-sensitive and loads files at startup, please set min replicas to 1 or more before the window begins so it does not cold start during the window.
- update · Oct 4, 9:00 AM EDT
Maintenance ended
What our checks recorded
We checked Cerebrium 24 times from Cerebrium's start time to its resolution, once every five minutes. Every one of them reached Cerebrium's site, so this incident did not show up as a failed page load. Our check loads the public site; it cannot see an API error or a broken feature. Median response time was 124 ms, against 138 ms over the 24 hours before. Cerebrium posted the incident at Oct 4, 7:00 AM EDT.
Updates are Cerebrium's own words from its status page (status: resolved); we never add an inferred cause.