A support case moves fast or slow depending on the first message. Give us the basics up front and the first reply works your problem. Leave them out and the first reply just asks for them. Here is what to gather.

## The essentials

We need these to find your cluster and reach you back:

- **Your name.**
- **Your organization identifier**, the workspace name we know you by.
- **The cluster name and its namespace**, both on the management cluster.

## The problem

- **What is broken**, in a sentence or two: what you expected, and what you got instead.
- **What changed just before it**: an upgrade, a scale, a config edit, a node added or removed. Most failures trace to the last change.
- **The exact error**: the `Reason` and `Message`, copied verbatim, and the object they sit on (a Cluster, Machine, Node, or PVC). See [how to read an event](/docs/hetzner/apalla/support/where-to-start#read-a-machine-or-cluster-event).
- **When and how wide**: when it started, in UTC, and whether it hit one node, one pool, or the whole cluster.

## If a node is involved

- **The Machine and Cluster events** for the affected objects, as text.
- **The log-collector bundle** from the node. See [collect logs with the log collector](/docs/hetzner/apalla/support/collect-logs-with-the-log-collector).

> [!TIP]
> If the fix will need Syself on the bare-metal server itself, prepare temporary access at the same time. See [grant Syself bare-metal access](/docs/hetzner/apalla/support/grant-syself-baremetal-access).

Send them on your channel. See [get support](/docs/hetzner/apalla/support/get-support).