MultiConnectMultiConnectDocs Website home
MultiConnectCloud › Server Deployments

16. Troubleshooting

The form refuses a step

Message What to do
Name must be between 3 and 60 characters. Shorten or lengthen the name
A deployment with this name already exists in the project. Choose another name
This project already has 20 deployments, which is the maximum. Delete a deployment, or use another project
Name the server binary — a Unity Linux build is named after your project, so there is no default. Enter the binary name, such as MyGameServer.x86_64, in Entry Point File
WebSocket needs TCP. A UDP-only server cannot be given a wss URL — nothing can terminate TLS for it. Tick TCP, or untick WebSocket
Port must be between 1024 and 65535. ... Use a port in that range, and listen on it in your server
MC_... is reserved — the platform injects the MC_* variables itself. Rename the variable. Read the platform's value instead (Section 10)
... is defined more than once. Remove the duplicate variable
Exactly one selected region must be marked primary. Star one region
Choose at least one region to host a master server. Tick a region under Master servers
Add at least one worker region. Workers run your matches, so a master needs them to place players. Click Add Worker Region
...: choose which master server these workers report to. Pick a master region in the worker row
...: minimum workers must be at least 1. Set Min to 1 or more, or use On demand
...: the ... plan allows at most N containers per region. Lower Max, or choose a larger plan
...: the ... plan allows at most N game servers at once per region. ... Lower that region's Most at once, or choose a larger plan
Start command contains a disallowed pattern '...'. ... Run one program with its arguments in Start Command. Move any setup into your server or into the zip
Start command contains a disallowed system command. ... Start your own server program, not a system tool
CPU scale-down threshold must be below the scale-up threshold. Lower the scale-down value. The same applies to memory
Autoscaling applies to worker servers only. ... Combined deployments do not autoscale. Add regions with Scale instead
Maximum instances must be between 1 and 50. Change Most running at once
Maximum instance lifetime must be between 10 and 240 minutes. Change Longest a server may run
Enter an https:// URL for webhook delivery. Use an https:// URL that is publicly reachable
The build is larger than the ... plan's N MB upload limit. Remove files from the zip, or choose a larger plan
Your balance is too low to run this configuration for even an hour (...). Top up, or reduce the container counts

Update, scale and plan changes

Message What to do
A ... operation is already in progress on this deployment (currently ...). Try again once it finishes. Wait for the current operation to complete
N previous generations are still draining and billing. ... Release a version under Previous Versions, or wait until one retires
Upload a worker build as well — the master and the workers are different programs. Choose Upload new build for both the master and the worker build
Server mode cannot be changed after creation. Create a new deployment instead. Mode, game servers and platform are fixed. Create a new deployment
Master regions are fixed after create. To run masters in different regions, create a new deployment. Leave the master regions as they are, or create a new deployment
Start the server before deploying to it. Click Start, then deploy
Configuration saved. It takes effect the next time this server is deployed. Not an error: the deployment is stopped or not yet deployed

Containers never become healthy

A redeploy that never becomes healthy is rolled back, and the previous generation keeps serving. Check:

Open Logs, choose Live tail, and read your server's output.

Clients cannot connect

Symptom Check
GetConnection answers "No healthy server is available right now." The deployment is running and its containers are Healthy on the Connections panel
A browser (WebGL) client cannot connect WebSocket is ticked, and the client uses the wss:// address, not the IP and port
Players land on an old version Your backend skips containers with retiresAt set, and acts on generation.retiring
A worker cannot reach its master The worker region's Reports to points at a master region, and the worker uses MC_MASTER_HOST and MC_MASTER_PORT

The lookup or instance API refuses a call

Message What to do
X-MC-Key, X-MC-Timestamp, X-MC-Nonce and X-MC-Signature are all required. Send all four headers
Timestamp is outside the 5-minute window. Check the clock on the calling machine. Sync the clock, and use UTC
Invalid signature. Sign exactly the path you call, and for instance calls hash exactly the body you send
Nonce already used. Every request needs a fresh one. Generate a new nonce for every request
Invalid or expired access key. Check the key on Access Keys
Unknown deployment. Check the deployment id, and attach the key to the deployment
This access key cannot manage instances. ... Create a key with Manage instances on, and attach it
This access key is marked as client-distributable ... Use an ordinary key for anything but lookups
Too many requests. Try again in a minute. Call less often: 60 requests a minute from each IP address

On-demand instances

Message What to do
This deployment does not use on-demand instances. The deployment uses Always ready game servers. Create an on-demand deployment
The deployment is stopped. Start it before creating instances. Click Start
The deployment has not finished its first deploy yet. Wait for the first deploy to complete
Instance limit reached: N of M running. Remove one, or raise maxInstances. Remove finished instances, or raise Most running at once
Region ...: N of M instances running. That region is at its Most at once. Raise it with Scale, or omit regionId
No host in ... has room for this plan right now. Try again shortly, or add another worker region
Balance too low: an instance can cost up to ... Top up, or ask for a shorter maxLifetimeMinutes
maxLifetimeMinutes must be between 1 and the deployment's maximum, N. Ask for a shorter time, or raise Longest a server may run
"..." is already set on the deployment; an instance cannot override it. Use a different name for the per-instance variable
Instance already created for this requestId. Not an error: you retried, and got the instance the first call started

For anything else, click Ask assistant on the deployment page, or see the MultiConnectCloud Guide for support.