My Stack

Fleeting

I kind of am very proud of my stack. The whole thing run on 3 raspberry pi 3B+ and an old laptop (bought in 2008).

• orchestration with nomad,
  • ingress with traefik,
  • networking with tinc,
  • servicemesh haproxy,
• file system with ipfs,
• clustering of ipfs with ipfscontroller (how I organize my static files with IPFS and org-roam, how do I store my files?),
• RPO : 4 hours (thanks to regular backups)
  • my scripts that import important data trigger a backup on the fly, so the RPO for those data is reduced to the time of the backup, about 10 minutes per GB.
• RTO : half a day, see ma stack permet une disponibilité moyenne,
• using manual chaos monkey,
  • unplugging from time to time a hard drive and erasing all of its content to simulate a need for changing one,
  • getting one raspberry pi out of the network for some time and resetting it to simulate a need for changing one

a.png

a.png

a.png

usb hubs to avoid

a.png

They lead to “over-current” indicated in dmesg, showing that the current is drawn from the raspberry pi instead of the hub.

Notes linking here

• configuration d’exporter prometheus
• how I organize my static files with IPFS and org-roam
• ma stack permet une disponibilité moyenne
• trying to setup retropie to play snes
• using 32 bits machines in my nomad cluster is harder and harder

Permalink