After the VM was rebuilt overnight, the booking site no longer responds

Riad Résa, a booking platform for riads and guesthouses in Marrakech, hosts its website www.riadresa.ma on a Compute Engine VM web-riadresa-01 (zone europe-west1-b, VPC vpc-riadresa, static external IP riadresa-web-ip).

Last night, the system disk was corrupted. Yassine, the on-call administrator, took a snapshot of the data disk, deleted the VM and then recreated it by hand with gcloud compute instances create (same name, same disk, same static IP). He left around 00:30, “everything was running”.

This morning, the site does not respond at all: browsers just spin, the day's bookings are blocked and travel agencies are calling. Yassine swears nginx is running.

You are on your workstation, with gcloud configured on the project riadresa-prod-4821 (default zone europe-west1-b). SSH access to the VM goes through IAP (rule allow-iap-ssh).