2026-10-02 13:34 UTCIncidents Detected

Akamai Edge DNS Issue

No internet-wide outage is in progress as of 13:30 UTC on October 2. Two items are worth knowing about: a new Akamai DNS propagation issue this morning, and a large Microsoft Azure networking incident on the evening of September 30 that our October 1 report did not cover.

**Akamai Edge DNS (investigating).** At 13:24 UTC Akamai opened an incident for Edge Delivery. Record changes made on customers' primary (master) DNS servers are not being reflected on Akamai Edge DNS. Sites that changed DNS records recently may still be serving old answers from Akamai's edge, while sites with unchanged records should not be affected. Akamai marks Content Delivery and Edge Delivery as degraded and has not yet given a scope or cause. This is at an early stage. ([Akamai status](akamaistatus.com))

**Microsoft Azure gateways (resolved, Sept 30 to Oct 1).** From 20:30 UTC on September 30, ExpressRoute Gateway and VPN Gateway connections degraded or dropped in about 18 regions, among them West US, North and West Europe, UK South, Japan West, Southeast Asia and South Africa North. Azure VMware Solution, Azure Firewall and Application Gateway were also hit. Gateways had mostly recovered by about 23:13 UTC, and Microsoft declared the incident mitigated around 02:15–03:30 UTC on October 1. Microsoft ties it to infrastructure OS servicing work and a change to a regional gateway management service that created more load than expected. That servicing has been paused. Hybrid-cloud and site-to-site VPN customers took the impact; public internet traffic to Azure was not broadly affected. ([Azure status history](azure.status.microsoft), [The Register](theregister.com))

**Other incidents from the past 24 hours (all resolved):**

**OpenAI**: login, signup and ads were failing from 18:46 to 19:39 UTC on October 1 (rated major), with full recovery after a mitigation. An earlier ChatGPT Space Pages error incident has also closed. ([OpenAI status](status.openai.com))

**Cloudflare**: increased errors for Durable Objects and R2 for about 20 minutes (20:59–21:18 UTC, Oct 1), and short network issues in Mumbai and Columbus, Ohio. The Madrid performance problem and the Workers build delays carried over from September 30 are now resolved. Two older, minor incidents remain open: Cloudflare One clients being challenged on some sites, and incorrect geolocation for some WARP users. ([Cloudflare status](cloudflarestatus.com))

**GitHub**: Actions job delays (14:47–17:56 UTC, Oct 1) and a short period of elevated request latency. ([GitHub status](githubstatus.com))

**Discord**: higher API error rates for about an hour on October 1. **Zoom**: a short degradation across Contact Center, Meetings and Webinars plus a Google Workspace add-on issue, both resolved.

**Ongoing minor issues:** Twilio is tracking carrier-specific messaging problems, including Silent Network Authentication failures on AT&T in the US, SMS delays to Telenet in Belgium and MMS delays to GCI in the US. Zoom Phone has an open issue with Polycom devices going offline. Akamai is still monitoring APAC origin connectivity tied to subsea cable repairs, which it expects to finish in early October.

**All clear:** AWS, Google Cloud, Azure DevOps, Anthropic, Atlassian, Slack, Vercel, Heroku, DigitalOcean, Cursor, Groq, ElevenLabs, Cohere and Stability AI report normal operation. We could not load Stripe's or Meta's status pages this run, so their status is unconfirmed, though we saw no outage reports for either. IODA flagged short, localized connectivity drops in a few countries over the past day, and no global BGP or DNS-root event has been reported.

**Bottom line:** the internet is up. If you run DNS through Akamai Edge DNS and changed records today, check that the changes have propagated.

← Home View all history

Akamai Edge DNS Issue

Last 30 Days