Some systems are having problems. Open the status page
Keep the gateway connected and your slash commands inside the three-second window, on a server that is entirely yours.
A Discord bot lives on the WebSocket connection it opens to the gateway. That connection expects a heartbeat at a fixed interval; miss it and the session drops and the bot shows as offline in the member list. Your library will try to Resume, but if the machine went to sleep there is no process left to resume from. Hosting a bot on your own computer means keeping a laptop tied to Discord's infrastructure.
Two things grow the resource requirement. First, cache: discord.js and discord.py hold the channels, members and messages of every guild the bot is in. The more guilds, the larger that table. discord.js lets you bound it with cacheWithLimits and sweepers, but not eliminate it. Second, voice: unless the source is already Opus, a music bot transcodes it with ffmpeg and then runs it through an opus encoder. That is CPU work a slash-command bot never does.
The third threshold is sharding. Discord's developer documentation states that a single shard supports at most 2,500 guilds and that apps past that number must enable sharding. Every shard is another WebSocket connection and another slice of memory, and the discord.js ShardingManager spawns them as separate processes. Size your plan for the day you shard, not for today.
Setup guide How to Install and Use Docker on a VDS: Beginner Guide
Your server runs in a data centre on redundant power and networking. Define the bot as a pm2 or systemd service and it comes back on its own after a crash. pm2's max_memory_restart will recycle a bot whose cache has grown past a threshold you set.
You connect as root. Install whichever Node.js version discord.js needs, Python for a discord.py stack, PostgreSQL or Redis for persistence. If you run Lavalink, it lives on the same machine as a separate JVM process and wants Java 17 or newer; it is the cleanest way to move audio encoding out of the bot process. Its default heap is 1 GB or 25% of system memory, whichever is lower, so size the plan for the bot process and the JVM together.
Your server comes with an IP that does not change, which matters when an external API whitelists by address or when your bot publishes its own interaction endpoint instead of using the gateway. Discord applies a global limit of 50 requests per second per bot; honouring the Retry-After header on a 429 keeps you clear of the invalid-request counter.
Past 2,500 guilds the Get Gateway Bot endpoint hands you a recommended shard count, and ShardingManager spawns those shards as separate processes. They can all live on one machine; the limit is their combined memory. Upgrade from the client panel and move to the new resources after a short restart.
The table assumes the bot runs alone on the server. If a database, a Lavalink node or a web panel shares the machine, pick one tier higher.
| Bot scale | Plan | Hardware | Monthly | Note | Order |
|---|---|---|---|---|---|
| 1-5 small bots · slash commands | ENT-1 | 2 vCPU · 2 GB DDR4 ECC · 30 GB NVMe | ₺160 | Discord publishes no memory figure and only advises keeping just the data you need in memory | Buy now |
| Music bot · ffmpeg and opus recommended | ENT-2 | 2 vCPU · 4 GB DDR4 ECC · 40 GB NVMe | ₺240 | No published figure for ffmpeg and opus; the encoding headroom is our own measurement | Buy now |
| Lavalink node · database | ENT-4 | 6 vCPU · 8 GB DDR4 ECC · 80 GB NVMe | ₺400 | Lavalink's official Docker example ships a 6 GB heap; the bot process and database sit next to it | Buy now |
| Sharded · 2,500+ guilds | ENT-5 | 8 vCPU · 10 GB DDR4 ECC · 100 GB NVMe | ₺480 | Discord requires sharding at 2,500 guilds and discord.js suggests 1,000 guilds per shard | Buy now |
Every plan includes a 10 Gbps uplink, DDR4 ECC memory and NVMe SSD storage. Upgrade any time.
The Xeon Enterprise line is more than enough for slash-command bots. Music bots that encode audio in the bot process are more comfortable on the Premium line.
A selection of technical articles from our blog.
Keep your Discord bot online around the clock on a VDS: installing Node.js, defining a service with pm2 or systemd, automatic restarts, log management and memory planning.
New to Linux servers? Learn how to connect over SSH, run the essential commands, harden the firewall and lock down your VDS on day one.
Install Docker on a Linux VDS, run your first containers, orchestrate services with Docker Compose and deploy a web app, step by step.
Common questions
Your server is ready 60 seconds after payment. Use code W10 for 10% off your first order.
Our support team is here to assist you with any questions or issues you may have. Contact us via phone, email, or visit our office during business hours.
