Troubleshooting & FAQs

Are you starting with empty disks?

The most common issue associated with the helm chart is persistent volume claim reuse. For example, if you deploy mygraph, delete this instance, and then redeploy a new, different mygraph, it will not get clean empty PVCs, but will reuse the old PVCs from the previous deployment.

A common problem is to install an instance with one password, which gets written to Neo4j’s system database. Later, when those PVCs are reused, an intial password cannot be set because it’s already been set.

Make sure to ensure the disks you’re starting with are empty to avoid file permissioning and other issues.

Neo4j’s method for installing plugins requires internet access. If you are in a locked-down networking environment, the automatic plugin installation may fail. In this case, you should set plugins: [] and use an initContainer to copy the appropriate plugins and extensions into your install manually.

This repo contains two helm charts; one for backup and the other to install a cluster or standalone instance. Ensure that when you run helm install you are pointing to the right directory.

Neo4j comes with a built-in debug.log file that is stored in the logs directory. This is the place to check when you run into trouble. Inspect this file and look for Exception error messages, and often the product logs will tell you what the issue is.

Yes.

When you use this repo, you’ll be pulling images for causal cluster, backup, and so on from the repository that’s provided. If you wish to host your own Docker images, you need to adjust the Makefile in tools that builds the docker images to point to your repo. You may then specify values to your install process that change the image and imageTag you are running. Check the repo’s top level README for information about building the tooling containers.

The Neo4j Community Site is a great place to ask for help.