Hacker News new | past | comments | ask | show | jobs | submit login

I completely agree.

I did a disaster recovery test of my services (all on docker) from scratch and without documentation.

I started by downloading the Debian ISO and went from there.

It took me about 60 minutes to be up and running, this includes DHCP from pihole and getting everything from borg backups.

I took notes to be faster/better next time but I lost them somehow :-|

Docker everything and you will be fine.

I used docker compose for years and recently moved to portainer because it is easier to keep track of the docker compose files and you have a lot of things readily available




I have “runbooks” for all my servers and software in Obsidian.

Every time I bootstrap it, I go through the instructions I wrote and improve/update if necessary.

Much simpler than Ansible


My last test was pre-Obsidian time (now I put everything in there). I will have to run one more time and, this time, make sure the steps are saved somewhere.

Since I basically install the OS, configure networking, install docker, install borg, recover from backup, start portainer, start the stacks - it would be overkill to use Ansible for that (not to mention hat I am not a big fan of Ansible after having tried for a year to make it reliably work on lossy/unstable networks -- IMHO salt is much better)




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: