Elastic favicon

Elastic Cloud outage on 2026-10-20

elastic.co
October 2026 20 Major incident

Elastic Cloud had two separate incidents on 20 October 2026. The longer one ran 15 hours 6 minutes and traced back to a GCP data centre fault in London, not to anything Elastic wrote itself. A second, shorter issue hit scaling for deployments running on Azure later the same day.

Started

06:51 UTC

Duration

Lasted 15h 6m

Source

IsDown

Next time

Elastic 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.

What happened?

The first incident started at 06:51 UTC when Elastic began investigating degraded availability for Elastic Cloud Hosted and Serverless deployments with capacity in GCP's europe-west2 (London) region. By 07:56 Elastic said its cloud provider had confirmed an infrastructure incident in a single availability zone, europe-west2-b, causing network connectivity loss and resource unavailability there. Deployments and projects with resources in that zone could be unavailable or degraded, and Elastic said it had no estimated recovery time from the provider. Updates through the day tracked slow, partial recovery: "early, limited signs of recovery" at 09:15, "slight signs of recovery" at 12:24, and an increasing number of deployments recovering by 14:15. Elastic marked the underlying infrastructure incident resolved at 17:27 but kept working on full service recovery, telling customers that some deployments might still need Support to step in individually. The incident was finally closed at 21:54, 15 hours 6 minutes after it began, with engineering confirming recovery of infrastructure in the affected zone. Elastic repeatedly advised customers whose plan changes or other operations failed during the window to retry them manually from the Cloud console. No user reports were filed during this window. The second incident began at 20:07 UTC, while the first was still unresolved, affecting the ability to scale existing Hosted deployments and Serverless projects running in Azure, plus delays creating new ones. Elastic identified the cause internally at 21:24, said it had mitigated it by 22:27, and confirmed scaling was working as expected again at 03:05 the next day, 6 hours 58 minutes after it started. Elastic's status page does not say what caused either the GCP zone failure or the Azure scaling fault beyond naming the cloud provider's infrastructure as the source of the first.

Learning

This was an infrastructure failure at the cloud layer underneath Elastic Cloud, not a fault in Elastic's own application code, and that distinction matters for anyone deciding where to point their monitoring. A check that only pings your own endpoints or your deployment's API would have shown the same degraded or unreachable state Elastic's customers saw, so in this case the symptom was visible either way, but the root cause sat one layer down, inside GCP's availability zone, where your own probes have no visibility at all. That is the general shape of this kind of failure: your product looks broken, your dashboards show errors, but the fix is entirely out of your hands and entirely out of Elastic's hands too, waiting on the cloud provider. Elastic's own history shows this is not rare, with 20 incidents in the last 90 days and a median duration of 5 hours 55 minutes, so extended waits on upstream recovery are a known pattern here, not a one-off. The Azure scaling fault on the same day shows the same provider-dependency risk can surface in a completely different cloud, on a completely different service path, within hours. Your monitoring covers your half, UptimeRobot now watches the provider's half too, so next time you know which side broke without guessing.

4.7
stars out of 5
284+ reviews on

Start monitoring in 30 seconds.

There's nothing to install. No credit card required. 50 monitors for free.