Nullpeak is a developer-first game host built around an API rather than a storefront: every server is a container you can define, and there is a documented REST API, a Terraform provider, crash webhooks and support for customer-supplied Docker images. Plans are hourly-billable so event and tournament servers can be created and destroyed from CI. Human support is business-hours email; the documentation is expected to do most of the work.
Share your experience to help others choose the right provider.
Strong platform overall, but the default filtering profile was dropping A2S query traffic, so for about a day our server showed as offline in the browser while being perfectly playable by direct connect. Support did eventually hand me a custom profile and it's been clean since. Would be five stars if the defaults matched the games they advertise support for.
I provision and destroy event servers straight from CI. The Terraform provider works properly, the REST docs have real examples instead of a schema dump, and crash webhooks fire fast enough that our on-call bot picks them up before players complain. First game host I've used that doesn't make me click through a portal like it's 2011.
Custom container images are a huge win, our ARK cluster runs the same image we test locally so no more 'works on my machine' arguments about the mod list. Docs are excellent for the API and thin for absolutely everything else, so week one was a lot of reading forum posts. Once it's set up it genuinely just runs.
Our node had a noisy neighbour that trashed tick rate two evenings running. I found it using their own metrics endpoint and sent them the graph with timestamps. Took roughly 48 hours to get moved to a different node and no credit was offered until I asked twice. The platform is good, the humans behind it are clearly stretched too thin.
Eight months in with a five-server modded MC network behind a proxy. Performance is good, the image system handles our custom jars without complaint, and my own monitoring has them at 99.9x uptime. Support is email-ish and business hours; twice I've waited overnight for a reply. Not a dealbreaker for me, but if you need someone at 3am on a Saturday, look elsewhere.
Strong platform overall, but the default filtering profile was dropping A2S query traffic, so for about a day our server showed as offline in the browser while being perfectly playable by direct connect. Support did eventually hand me a custom profile and it's been clean since. Would be five stars if the defaults matched the games they advertise support for.
I provision and destroy event servers straight from CI. The Terraform provider works properly, the REST docs have real examples instead of a schema dump, and crash webhooks fire fast enough that our on-call bot picks them up before players complain. First game host I've used that doesn't make me click through a portal like it's 2011.
Custom container images are a huge win, our ARK cluster runs the same image we test locally so no more 'works on my machine' arguments about the mod list. Docs are excellent for the API and thin for absolutely everything else, so week one was a lot of reading forum posts. Once it's set up it genuinely just runs.
Our node had a noisy neighbour that trashed tick rate two evenings running. I found it using their own metrics endpoint and sent them the graph with timestamps. Took roughly 48 hours to get moved to a different node and no credit was offered until I asked twice. The platform is good, the humans behind it are clearly stretched too thin.
Eight months in with a five-server modded MC network behind a proxy. Performance is good, the image system handles our custom jars without complaint, and my own monitoring has them at 99.9x uptime. Support is email-ish and business hours; twice I've waited overnight for a reply. Not a dealbreaker for me, but if you need someone at 3am on a Saturday, look elsewhere.