Running sky launch on a new GCP project (without running sky check first) fails because required APIs may not be enabled. This is expected.
However, at a quick glance, the error message shown to user seems to suggest going to the cloud console as a solution instead of sky check.
Even though sky check is proposed as a solution, it gets buried 7+ lines above the last line. If the user does not read the whole stack trace carefully, they might think the cloud console URL is the solution since 1) it's printed later and 2) is long, thus easily noticed.
I 01-24 22:13:07 cloud_vm_ray_backend.py:3143] Creating a new cluster: "vpc" [1x GCP(n2-standard-8)].
I 01-24 22:13:07 cloud_vm_ray_backend.py:3143] Tip: to reuse an existing cluster, specify --cluster (-c). Run `sky st
atus` to see existing clusters.
I 01-24 22:13:57 cloud_vm_ray_backend.py:1081] To view detailed progress: tail -n100 -f /home/romilb/sky_logs/sky-202
3-01-24-22-12-33-267292/provision.log
E 01-24 22:14:19 authentication.py:172] Certain GCP APIs are disabled for the GCP project romil-sky-exp.
E 01-24 22:14:19 authentication.py:174] Enable them by running: sky check
E 01-24 22:14:19 authentication.py:176] Details:
E 01-24 22:14:19 authentication.py:177] Compute Engine API has not been used in project 840775419424 before or it is disabled. Enable it by visiting
E 01-24 22:14:19 authentication.py:177] https://console.developers.google.com/apis/api/compute.googleapis.com/overview?project=840775419424
E 01-24 22:14:19 authentication.py:177] then retry. If you enabled this API recently, wait a few minutes for the action to propagate to our systems and retry.
Clusters
NAME LAUNCHED RESOURCES STATUS AUTOSTOP COMMAND
upload 1 day ago 1x GCP(n2-standard-8) UP - sky launch --cloud gcp -c...
Local clusters:
NAME USER HEAD_IP RESOURCES COMMAND
localclus - - - -
