Longhorn installs easily when the cluster is ready for it. The problem is that clusters are often not ready for it.
This note was really about that preparation step.
1. Install the Node Dependencies
Before touching Helm, make sure the nodes have the packages Longhorn expects:
| |
Then enable the iSCSI service on the nodes:
| |
That sounds small, but it is exactly the sort of small missing dependency that wastes time later.
2. Run the Longhorn Environment Check
The most useful command in the note was this one:
| |
I like this step because it catches the cluster-level assumptions before the actual chart install.
In the original note, the check exposed a few warnings:
- missing or unreachable routing for the control plane
- older kernel versions
multipathdrunning on the nodes
That is exactly the kind of feedback you want before you start blaming the storage system itself.
3. Fix the Basic Network Reachability First
One of the quick fixes from the note was simply restoring the expected default route before re-running the check:
| |
That is a good reminder that not every “storage install problem” is actually a storage problem.
4. Install Longhorn Only After the Checks Look Reasonable
Once the environment check was clean enough, the chart install was straightforward:
| |
Then watch the namespace:
| |
Closing Thought
The biggest value in this note is not the helm install line. It is the reminder to treat Longhorn as a node-readiness problem first and a chart-install problem second.
That order saves time.