I’m not a registered company, but a solo dev who creates things for joy and ends up getting paid for it.
Legacy projects: Plain HTML frontend with minimal vanilla JS for small progressive enhancements, PHP (custom framework) backend, Postgres, deployed directly on rented bare metal hosts, Salt for configuration
New projects: HyperApp JS frontend, Python (aiohttp) or Rust (warp) backends, Postgres, deployed in docker containers on rented bare metal hosts, Ansible for configuration
Notes:
switched from salt-server to salt-ssh after there was a salt-server exploit which owned my whole fleet. Switched from salt-ssh to ansible-ssh after salt-ssh was broken on osx for 14 months (and counting…)
Backups done with each project having a cronjob dumping a snapshot into a folder, then SyncThing copies those onto a remote backup host, with the backup host configured to keep old versions of files
Debian stable for the hosts, language-specific images for containers. I wonder if my bare-metal OS runs basically only sshd and docker, is there a better choice than Debian? It works fine, just seems overkill...
Monitoring done with telegraf + influxdb. Previously I had grafana in front. Recently moved to influxdb 2.0 with built-in dashboards, which is ok, but grafana was better (more graph settings (influx doesn’t even do log-scale y axis??); more flexible alarming; graphing and alarming were integrated where influx has them separate) - considering putting grafana back even if it means re-creating all the dashboards and alarm rules for a second time...
Legacy projects: Plain HTML frontend with minimal vanilla JS for small progressive enhancements, PHP (custom framework) backend, Postgres, deployed directly on rented bare metal hosts, Salt for configuration
New projects: HyperApp JS frontend, Python (aiohttp) or Rust (warp) backends, Postgres, deployed in docker containers on rented bare metal hosts, Ansible for configuration
Notes:
switched from salt-server to salt-ssh after there was a salt-server exploit which owned my whole fleet. Switched from salt-ssh to ansible-ssh after salt-ssh was broken on osx for 14 months (and counting…)
Backups done with each project having a cronjob dumping a snapshot into a folder, then SyncThing copies those onto a remote backup host, with the backup host configured to keep old versions of files
Debian stable for the hosts, language-specific images for containers. I wonder if my bare-metal OS runs basically only sshd and docker, is there a better choice than Debian? It works fine, just seems overkill...
Monitoring done with telegraf + influxdb. Previously I had grafana in front. Recently moved to influxdb 2.0 with built-in dashboards, which is ok, but grafana was better (more graph settings (influx doesn’t even do log-scale y axis??); more flexible alarming; graphing and alarming were integrated where influx has them separate) - considering putting grafana back even if it means re-creating all the dashboards and alarm rules for a second time...