Troubleshoot instances
Fixes for instances that will not start, cannot be reached, refuse your password or fail a snapshot or backup.
Find the message or the symptom you see, then follow the fix. Each entry links to the page that covers the task in full.
Ordering and setup
Section titled Ordering and setup- An instance you ordered is not on the Cloud Compute page
- The page lists an instance once we have built it. Open Services and find the order.
Pendingmeans it waits for payment or for setup to start,Provisioningmeans we are building it, andFailedmeans setup did not finish. Manage your services explains each status. For a failed order, contact support with its service number. - No address on the instance
- The instance has no public IP address. If you ordered the included address, we could not attach it, and it is on your Floating IPs page: attach it to the instance, then restart the instance from the header.
- You do not know the instance's password
- We set a password for the image's default user when we build the instance, but we never show it. Sign in with the SSH key you added, or reset the password on the Access tab.
Power and boot
Section titled Power and bootThis instance is suspended- We suspended the service behind the instance, usually for an unpaid invoice. Select Open Billing in the alert and pay the open invoice. Failed payments and suspended services explains suspension.
Cannot perform power actions on a suspended instance- The same cause as above. Pay the open invoice, then try again.
Instance is migrating- We are moving the instance to another host. Wait until the move finishes and try again.
- Stop or Restart does nothing
- The operating system ignored the request to shut down. Stop powers the instance off after 60 seconds anyway. If a restart hangs, select Kill… in the header's Instance actions menu, then Start. Start, stop and restart an instance covers each action.
- The instance boots into rescue media every time
- The instance is still in rescue mode, and the header shows a
Rescuebadge. Leave rescue mode on the Rescue tab. - The console shows a boot error, or the instance never reaches a sign-in prompt
The operating system on the disk cannot start. Choose one of these fixes:
- Repair it in rescue mode to keep the data.
- Roll back to a snapshot taken before the problem.
- Restore a backup.
- Rebuild the instance, which erases the disk.
The instance is powered off- The console needs a running instance. Select Start in the header.
Could not connect to the console- The console session expired before it connected, or the host did not answer. Select Try again. Use an instance's console lists the other console messages.
Connecting
Section titled ConnectingConnection timed outorNo route to host- The instance is stopped, has no public address, or has not picked up an address you attached. Start or restart it from the dashboard header, which boots it with its current network settings. If the address is a floating IP, check the rules on its Shield tab.
Connection refused- The address answers, but no SSH server listens on port 22. Sign in on the display console and start the SSH service, or check which port it uses.
Permission denied (publickey)- The instance does not hold the key you used, or you signed in as the wrong user. The Access tab shows the user. Keys you save there apply only after you restart the instance from the header, and a rebuild does not put them back. Connect to an instance over SSH has the details.
- The instance refuses your password
- The password inside the instance is not the one you have. Reset it on the Access tab while the instance runs. If the reset fails, set a new password from rescue mode.
WARNING: REMOTE HOST IDENTIFICATION HAS CHANGED!- The instance has new SSH host keys. A rebuild creates new ones, and cloud-init usually does too on the first boot after you change the instance's keys, password, hostname or network settings. If you made one of those changes, run
ssh-keygen -Rwith the instance's address, such asssh-keygen -R 203.0.113.10, and connect again. - A second floating IP does not answer
- After a rebuild or a cloud-init regeneration, the instance sets up only its primary address. A restore brings back the addresses the instance had when we made the backup. In each case, detach the other address and attach it again, then restart the instance from the header.
Snapshots and backups
Section titled Snapshots and backupsSnapshot limit reached- The instance holds 8 snapshots. Delete one before you take another.
- A snapshot fails as soon as you take it
- A name longer than 37 characters is one cause. Choose a shorter name. Take and roll back instance snapshots lists the other causes.
Backup limit of 16 reached- Delete a backup you no longer need, then create the new one.
A backup or restore is already in progress on this instance- The instance runs one backup job at a time. Wait until the table shows the job has finished.
Monthly traffic quota exceeded; backups are blocked until the next period- The instance has used its traffic allowance for the month. New backups start again on the 1st. How instance traffic is counted explains the allowance.
Too many requests for this action. Please wait and try again.- You sent the same action too often for this instance. Cloud Compute limits lists each limit and its window.
Size and billing
Section titled Size and billing- The instance needs more vCPU, memory, disk or traffic
- An instance keeps the size you ordered. Get a different size explains how to move to a new instance.
Over the plan allowance- The instance has used more traffic this month than its plan includes. Its traffic still flows, but we make no new backups until the next month.
Errors from the API
Section titled Errors from the API401- The access token is missing or has expired. Authentication explains how to get a new one.
404withCloud Compute instance not found- The UUID is wrong, the instance belongs to another account, or it has ended.
403withInstance is suspended- The service is suspended. Some actions answer
Cannot perform power actions on a suspended instanceorCannot rebuild a suspended instanceinstead. Pay the open invoice first. 409withInstance is migrating- We are moving the instance to another host. Try again once the move has finished.
429- The action is over its limit for this instance. Wait for the number of seconds in
Retry-After. Cloud Compute limits shows the response.
Errors explains the shape of every error response.
Get help
Section titled Get helpIf none of these fixes the problem, contact support from the dashboard. Include:
- the instance ID from the Identifiers card on the Settings tab, and the Service ID from the Billing tab
- what you did and when
- the message you saw, word for word