A hosted custom Discord bot's uptime depends entirely on the infrastructure behind it, and with Noxyr's Custom Bot, that's a set of hosting nodes that pick a host for your bot automatically, keep checking on it every few seconds, and move it without you ever touching a server yourself. You paste a token; you don't provision, patch, or restart anything.
You don't manage a server, Noxyr's nodes do
Each node is one process that hosts every custom bot assigned to it, checking in on a short cycle to start bots that should be running, stop ones that shouldn't, and record a heartbeat for the website to read. That's the entire job the infrastructure does that you'd otherwise have to do yourself on a VPS: keep the process alive, watch for it dying, and log in again when Discord disconnects it for a reason that isn't your fault.
How a host gets picked when you set up
Automatic is the default and it's the one most people should leave selected. It always resolves to whichever node currently has room, and nodes fill in a fixed order, one is used up before the next one starts taking new bots, rather than spreading load evenly and leaving every node half full. Picking a specific host manually still routes you to a real, working node behind the scenes: if the one you picked happens to be offline or full right then, the best available one is used instead of the setup simply failing.
What makes a node "full"
Two checks decide it, not one. A node stops taking new bots once it hits its own bot-count limit, and separately, once one more bot would push its measured memory use close to its ceiling. The second check exists because a headcount limit alone doesn't account for busier servers using more memory per bot than quiet ones; a node that looks like it has room by bot count can still be too close to running out of memory to safely take another. Whichever limit is hit first, new setups roll onto the next node automatically.
Moving your bot to a different host
Change Node on the Custom Bot page moves an already-running bot elsewhere. Every move is a fresh Discord login, and Discord limits how often any one bot can do that, so there's a five-minute cooldown between moves. A real move briefly restarts the bot while it reconnects on the new node, your settings, tickets, catalog and everything else are untouched, since that all lives in the shared database, not on whichever node happens to be running the process.
What happens if a node has a problem
If a bot's heartbeat goes stale, its node stopped hearing back from it, Noxyr's main bot steps back in and covers that server again automatically, so members are never left with genuinely zero bot answering. Once the node or the login recovers, the custom bot picks the server back up on its own on the next check-in. Nothing about this needs a support ticket in the ordinary case; it's the same self-healing loop that starts your bot the first time you set it up.
Why nodes fill in a fixed order instead of spreading out
It would be easy to assume load balances evenly across every node, a bit on this one, a bit on that one. Noxyr does the opposite on purpose: one node fills up completely before the next one takes a single new bot. That keeps the number of nodes actually running at any moment as small as it can be for the current load, rather than lightly loading several at once, which matters for how efficiently the underlying infrastructure gets used. It has no effect on your bot's behavior either way; the sync loop and heartbeat work identically regardless of which node ends up running it.
What the numbers on the dashboard actually mean
Ping shown against a host is a real measurement, not a guess, each node periodically times a request to Discord's own gateway endpoint and reports the average. A node's online/offline state is based on how recently it last checked in, a node that's gone quiet for a short while is treated as unavailable rather than assumed to still be fine, which is also what triggers Noxyr's main bot to step back in and cover a server whose custom bot has gone quiet.
Uptime vs a paid VPS
Running your own VPS gives you full control over the machine, and full responsibility for noticing when it goes down. A hosted custom bot trades that control for someone else's job being to notice first. Neither is strictly better in every case, see our full comparison in custom bot vs self-hosting on a VPS if you're deciding between the two.
Common questions
How do I check if my bot is online right now? The Custom Bot page on the dashboard shows live status, not just whether it's switched on.
Do I lose settings when my bot moves to a different node? No. Settings live in the shared database, independent of which node's process happens to be running the bot.
Can I pick a specific node instead of Automatic? Yes, but it always lands on a real, working host even if the one you picked is full or offline at that moment.
How often does the status on the dashboard refresh? Every few seconds; a node that stops hearing from your bot is noticed quickly, not after a long delay.
Is downtime billed or penalized in any way? No. There's no payment involved at all, uptime issues don't cost you anything beyond the bot being briefly unavailable while it reconnects.
Full setup requirements are in the Custom Bot docs, and the exact node-picking and Change Node mechanics are in the custom bot hosting docs.