Reinstalling the OS: What are the correct steps to bring my nodes back online on Debian 13?

Node Operators troubleshooting
1

Hi everyone,

I recently had a hardware issue with my NUC: the NVMe drive containing the operating system failed completely. However, the disks where my Storj nodes store their data (one 4TB disk and one 16TB disk) are intact, including all databases and stored data.

I have already reinstalled the operating system, but this time I used Debian 13 instead of Ubuntu. The data disks are mounted correctly, and I also have the identity folders for each node.

Before starting the containers again, I want to make sure I follow the correct procedure to avoid any reputation loss or data issues.

My main questions are:

  1. What is the recommended procedure when only the OS disk fails but the node data disks remain intact?
  2. Are there any additional steps I should take when using Debian instead of Ubuntu?
  3. Can I simply run the Docker container again, pointing to the existing identity and data folders?
  4. Should I check or repair anything in the databases before starting the node?
  5. Is there anything else I should verify to ensure the node resumes normally?

Additionally, since I am a medium‑level user and not an expert, I would really appreciate it if someone could provide the exact commands or steps needed to correctly reinstall and start the nodes again on Debian.

I want to avoid making a mistake that could affect the reputation of my nodes, so any guidance or experience from users who have gone through this situation would be greatly appreciated.

Thanks in advance.

2

Everything is pretty simple, if your identity and data on a dedicated disk, separated from the OS drive:

  1. Install OS
  2. Update OS
  3. Install Docker
  4. Maybe update permissions/owner of the identity and data folders to match a new one.
  5. Run the container pointed to the existing identity and data.

I would like to recommend to use a docker-compose file, because to run a node(s) your command would be as simple as:

docker compose up -d

and all done - the node(s), the watchtower and all are configured right in the docker-compose.yaml file, up and running.

3

This “docker compose” thingy needs something installed extra? Like a “compose” extension or something? Or is part of the docker?
If I run the node with the usual docker run… can I switch to compose without installing anything? Just make a compose.yaml and run the docker compose command?

4

It should be a part of the latest docker package, but I could be wrong and it still requires to install it additionally.
The plugin gives you a command docker-compose, while docker gives you docker compose (without a dash).

5

I’ve found Docker nodes to be very portable. Move their directories to a new place, adjust any “volumes:” paths in docker-compose.yml… then start them up! The only extra steps are usually to remember to move the “:/app/dbs” location too (if you’ve placed the databases on SSD or something)

1 Like
6

Or don’t adjust the volumes path at all, because you’ve already set the root directory of each node to have docker-compose.yml, data/ and identity/, and you’ve just set

   volumes:
      - ./identity:/app/identity   
      - ./data:/app/config

My nodes are moved around all the time. Easy as pie

2 Likes
7

You can’t easy move a pie. Usualy the pie dosen’t reach the destination in full. :face_savoring_food:

7 Likes
8

I bet you hate this one little naughty maneuver here

image
image1300Ă—956 128 KB

1 Like