Tink vs node-problem-detector
node-problem-detector is a Kubernetes DaemonSet. It watches kernel logs and runtime sockets, then paints Node conditions so the scheduler can taint a bad Node. That is not a mechanic. Tink installs in one command, watches the Linux server, explains what is wrong, and helps you fix it.
NPD reports Node conditions. Tink answers “why is this VPS sick?”
Accidental sysadmins land on node-problem-detector because every Kubernetes node-health diagram mentions it. Then the volume fills, nginx dies, and the Node stays Ready — the VPS was never a Node. A filling /var is a Tink warning; it is not a KernelOops unless you already run NPD on that box.
Tink is the other job: detect the issue on the machine, say why it happened in plain English, propose the command, and run it only after you approve. Keep node-problem-detector if you already operate Kubernetes and need Node conditions. Use Tink if you run Linux servers and want a mechanic.
| Feature | Tink | node-problem-detector |
|---|---|---|
| Setup time | 30 seconds (one curl | sh command) | Hours — a cluster, DaemonSet, problem config, and still no host mechanic until Prometheus and a taint policy exist |
| What you get | Working monitoring, diagnosis, and approved fixes | Node conditions plus events. The box is still yours to diagnose and SSH |
| Pricing | Free (Scout) / $9 / $29 per machine per month | Free software (Apache 2.0). You still pay for Kubernetes, a metrics stack, and the engineer who owns NPD rules |
| Hidden costs | None — fully managed | Custom problem plugins, kernel-log parsers, taint controllers, and a second tool that pages you when /var fills |
| Monitoring approach | Agent on the server — CPU, disk, services, logs, certs, ports | Node conditions — NPD is how Kubernetes learns KernelDeadlock, not a host mechanic |
| Configuration | None after install — heuristics and AI | config/kernel-monitor.json, custom plugins, and which systemd units NPD is allowed to watch |
| Plain-English diagnosis | Yes — AI explains root cause, impact, and fix | No — KernelOops is a condition. You still SSH or kubectl describe node to learn why |
| Fix execution | Proposes and executes approved commands with an audit trail | Conditions only — NPD cannot free disk, restart nginx, or rotate a cert on a VPS |
| Alerting | Built-in across 8 channels | None — kube-state-metrics or a taint controller must notice the condition before anyone is paged |
| Predictive alerts | Yes — disk fills in ~6 days, memory and CPU trends | Never — NPD fires after the kernel already oopsed. It does not forecast host fill |
| SSH brute-force detection | Built-in — parses auth.log every scan | Never — NPD does not read auth.log, journald SSH, or host firewall state |
| Machine offline detection | Agent presence monitoring with multi-channel alerts | A silent NPD is a missing DaemonSet pod — only if the API server still answers |
| Public status page | Shareable URL with 90-day history | None — node-problem-detector has no customer-facing status page |
| Weekly fleet digest | Automated Monday digest + daily brief when issues are open | None — a KernelDeadlock condition is not a plain-English fleet narrative |
| On-call tracking | Built-in /oncall command + incident acknowledgment | Not included — wire Node conditions into PagerDuty or another incident tool |
| Conversation interface | Telegram, WhatsApp, web dashboard, CLI | kubectl describe node. No mechanic you text when /var filled on a VPS that is not a Node |
| Learning curve | None — works after install | Steep — Kubernetes, DaemonSets, problem configs, taints, and which Node condition actually means the box is sick |
| Best for | Freelancers, small teams, accidental sysadmins (1-50 Linux servers) | Teams that already run Kubernetes and need Node conditions, not a sick-VPS mechanic |
When node-problem-detector is the right choice
Keep node-problem-detector when you already want that exact job:
- Node conditions — NPD is how Kubernetes learns KernelDeadlock and DockerHung. You need it to taint Nodes, not to watch a quiet VPS.
- An existing cluster — DaemonSet, problem config, and a taint controller already exist, and someone owns kernel-log parsers.
- Scheduler questions — NotReady, taints, and eviction matter more than host nginx, certs, and SSH.
- Hosts that only exist as Nodes — you already run Kubernetes, not a sick-box SSH loop.
The real cost of “just run NPD”
node-problem-detector is in every Kubernetes node-health diagram. A working ops loop for five Linux boxes is not:
- A KernelOops condition still means you SSH in and change the box by hand
- Host fill, nginx, certs, and SSH brute-force never appear unless they show up as a problem plugin you already wrote
- A filling /var is a Tink disk warning on every scan — NPD will stay quiet until the kernel already failed
- A single VPS with no cluster still needs a mechanic, not a DaemonSet
For a 5-server team, Tink Mechanic at $45/month is cheaper than node-problem-detector plus the engineer who keeps problem configs and taint controllers alive, and every scan includes a diagnosis a quiet Node condition will not type.
No DaemonSet homework. No silent host. One command install.
Start MechanicAlso compare: Tink vs kubelet · Tink vs kube-state-metrics · Tink vs Prometheus