Server stuck pending, or node not assigned

A new deployment starts as pending. Pending means the deployment exists in your account but the work of placing it and installing it is not finished yet.

What normally happens

1. The panel reserves a slot and the resources from your plan.
2. It picks a node with room for it.
3. It registers the deployment on that node and uploads your files, or clones your repository.
4. The install runs, then the start command runs.

Steps 1 to 3 take seconds. Step 4 is what usually takes time: installing dependencies on a small CPU share can take several minutes, and a large repository takes longer to clone. The console shows all of it live, so open the console instead of watching the state badge.

"Servers are full"

If a deployment is refused with a message about capacity, the platform could not place it. That happens when every node that accepts new deployments is at its limit for your plan's tier.

What to do:

• Try again a little later. Capacity is freed continuously.
• A paid plan has access to headroom a free deployment does not, so upgrading usually places immediately.

The node did not answer

Sometimes the panel shows blank resource usage, or a console that will not connect, and says the state is unknown. The panel deliberately says "unknown" rather than guessing "offline", because your app may well still be running.

This is almost always temporary. Refresh after a minute. Your files are untouched.

A deployment that never leaves pending

If several minutes have passed, the console is empty, and nothing at all appears:

1. Refresh the page once. The state badge is polled, and a stale tab can lag.
2. Open the console. Any output at all means the install is running and you just need to wait. Do not restart or kill it while it is installing: that is the fastest way to end up with a half installed project.
3. If the console is truly empty, the deployment was never placed. Delete it and create it again. Deleting is instant and frees the slot and the resources right away.
4. If recreating gives the same result, open a ticket. Some cases need staff to re register the deployment on a node, which cannot be done from your side.

Do not create the same deployment over and over: each attempt holds a slot from your plan until you delete it, and you will hit your slot limit instead of getting further.

What to send in a ticket

• the deployment ID, from the URL of its page
• when you created it
• whether the console showed anything at all
• the exact message, if the panel refused the deployment