Google Cloud's London Zone Went Down at 06:08 UTC. Its Status Page Waited 2 Hours 27 Minutes

A power event took out the europe-west2-b zone in Google Cloud's London region at 06:08 UTC on Tuesday, October 20, and the first public status post did not appear until 08:35 UTC, according to Google's own incident record. The record lists the incident as ending at 14:00 UTC, which makes it 7 hours and 52 minutes long.

The cause went unnamed for five hours. The 11:12 UTC update says: "A power event caused an outage within europe-west2-b zone." Until then every update said Google was investigating errors and intermittent network failures, with no workaround.

What the incident record says, update by update

The record holds seven updates, and Google's status tooling marks six of them as a service disruption. Only the last one, at 14:03:41 UTC, flips the status to available.

Time (UTC)	What Google posted
08:35	Investigating errors and network failures in europe-west2. No workaround.
09:24	Still investigating. Looker Studio not affected.
11:00	Still investigating.
11:12	Power event in europe-west2-b named as the cause.
11:58	Services recovering, existing workloads first. New workloads may not be creatable.
12:44	Most workloads recovered as of 05:10 PDT. Migration to another zone suggested.
14:03	Mitigated for most customers. "No further updates will be provided here."

The 11:58 update is the one that matters to anyone who tried to recover by hand. Google said it was "prioritizing the recovery of existing workloads" and that customers "may not be able to yet create new workloads". A team that tried to rebuild in the same zone would have hit that wall.

1. First status post147 min
2. Cause named304 min
3. Most workloads recovered362 min
4. Marked available475 min

Source: Google Cloud incident record Xs2XTHzAVC9YnafnPyzj, accessed 2026-10-21

The 147 minutes before the first post are the part customers could not plan around. A runbook that says "check the status page before failing over" would have sent an on-call engineer to a page that was silent for two and a half hours.

Who noticed first

The Register's EMEA Editor, Paul Kunert, reported that the zone is one of three in the London region. He listed seven Google services Google had flagged, among them Google Kubernetes Engine, Cloud Run, Google App Engine and Cloud Filestore.

The downstream list in that report comes from Downdetector, not from Google. It names Lloyds Bank, HSBC, Royal Mail, Rightmove, DPD, BT, Screwfix and Revolut. Revolut said that "all our customers' money and data are safe."

Neither Google's record nor The Register says how many customers were running in europe-west2-b alone. Downdetector counts user reports, so the list shows who was noticed, not who was down.

What Google has and has not promised

The closing update says engineers are still working with "a very small number of customers" and that Google will "publish an analysis of this incident once we have completed our internal investigation". It gives no date for that analysis. The word "power" is all the record offers on what failed: no utility fault, no generator, no cooling.

Google's own workaround text, posted at 12:44 UTC, was to migrate to another zone or region "if feasible". That advice is useful only to a team that already had a second zone provisioned.

The analysis, once published, should say whether the 147-minute gap was detection or publishing. Until it does, a status page timestamp is the time Google wrote the post, not the time the zone failed. Compare it with Jepsen's nine findings on Orkes Conductor for the kind of failure account that names a mechanism, and with the TorchTPU 0.2.0 release for another Google infrastructure product with open questions on its page.