Compute Troubleshooting

When the console refuses something, it says why. This page groups the messages you are most likely to meet, what they mean, and what to do. Product-specific refusals are also listed on each product's page.

I don't see a Compute product, or it says "No access"

Compute is open to every account. If you are working in someone else's account, the sidebar only shows the products you were given access to, and a page you weren't given says No access to followed by the product. Ask the account owner or a team manager for access, or switch back to your own account from the profile menu. An action you can see but not use needs a higher level: the page says which. See Shared accounts.

"Not enough balance for this launch"

Your balance must cover 24 hours of everything you run, plus what you are launching. The message tells you how much to add. Top up from Billing, then launch again. See the 24-hour balance requirement.

If the message says your balance doesn't cover 24 hours of the compute you already run, you have to top up before you can launch anything, or delete what you no longer need.

"Your balance could not be read just now" means we couldn't check your balance. Nothing was launched. Try again in a moment.

"This launch is over your compute quota"

The launch would take your account over one of its limits. The message names each limit, how much you use and how much the launch adds. Delete what you no longer need, or ask support to raise your limits. Stopped VMs count too. See Quotas.

"No verified server in this region" or "No server can take this size"

There is no server with room for what you asked, where you asked for it.

• Pick another region, or Automatic (any region) for a VM.
• Pick a smaller size.
• Try again later: capacity changes.

A Kubernetes cluster always needs a region, because its three masters run in one country.

"No public IPv4 address is free right now"

All public addresses are in use. Launch without one and attach it later from the VM's page, or try again later. You can still reach the VM from the browser with the Console tab.

A VM launch failed

If a VM ends up Failed, its page explains why:

Explanation	What to do
No verified server in this region right now.	Try another region.
No compute node can take this VM right now.	This is capacity, not your configuration. Try again shortly, or pick a smaller size.
No compute node has room for this size right now.	Try a smaller size, or again shortly.
That image is not available on the compute nodes.	Pick another image.
The compute node did not answer in time.	Nothing was charged. Try again.
The node could not prove it is running confidential hardware.	The launch was refused to protect you. Try again.

A VM is only billed from the first time it runs. Delete a failed VM to remove it from your list.

I can't connect over SSH

See SSH troubleshooting. In short: use the image's username, not root; check that the VM has a public IPv4 and a firewall rule allowing port 22 from your address; or use the browser Console tab.

A power action is refused: "only a running VM can be powered"

The VM hasn't finished provisioning, or is being deleted. Wait until it is Running. If a resize or a restore is running, wait for it to finish.

"This VM is being resized; delete it once the resize has finished"

A VM can't be deleted during a resize. Wait for the resize to end, then delete it.

"Your account is being deleted"

New resources can't be created while your account deletion is pending. Cancel the deletion first if you want to keep using Compute.

Error codes in API responses

If you call the API directly, refusals come with an HTTP status and, for most of them, a machine-readable code. It is in code, or in error with a sentence in detail for the public IPv4, firewall, resize and backup endpoints. The most common:

Code	HTTP status	Meaning
compute_quota	409	Over a quota. The body lists exceeded, limits and used.
(no code)	402	Balance below 24 hours of compute. The body has required_credits and balance, in dollars.
(no code)	503	The balance couldn't be read. Retry.
compute_arrears	402	Unpaid compute usage; services are stopped. Top up.
compute_arrears_suspended	409	Paid; the database or cluster restarts within the hour.
account_deletion_pending	409	The account is being deleted.
no-free-public-ip	409	No public IPv4 is free.
public-ip-unavailable	503	Public IPv4 can't be offered right now.
vm-not-live	409	The VM must be running for this action.
firewall-changed	409	The firewall rules changed since you read them. Read them again.
resize-in-flight	409	The VM is being resized.
resize-cooldown	429	The VM was resized a moment ago. retry_after_s says when to retry.
job-in-flight	409	Another operation is running on the VM.
no-operator	409	The cluster can't run node operations. Create a new cluster to use worker pools.

In an account you are a member of, refusals of the shared account itself are a 403 with codes such as not_a_member or insufficient_grant. They are listed in Shared accounts.