Skip to content

Backup docker-compose deployments and restore prod data into canary

Benjamin Bock requested to merge 676-create-backups into master

This MR closes most of #676:

  • creates backup of the non-canary systems (configurable via CI variable in deployment step)

    • SQL Dump and upload to storage box
    • tar storage files and upload to storage box

  • restores latest backup into non-canary systems from their prod instance right before deploying

  • keeps any backup, both created before deployment as well as from cron jobs (need to clean up manually for now, amount of data is still tiny so this shouldn't be an issue)

  • Neues Pendant für cap production postgres:replicate:

    • Read-only Storage Box Login (SSH Public Key des Entwicklers muss dort hinterlegt werden)
    • ssh-config muss wie im storage-box abschnitt konfiguriert werden
    • restore-script braucht noch ein bisschen Finetuning für lokalen Betrieb (rechte in Non-Docker etc) → followup ticket
    • Zum Ziehen des aktuellsten Alpha-Backups: ./docker/operations/backups.sh restore samarbeid-alpha
    • Zum Ziehen des aktuellsten Claudia-Backups: ./docker/operations/backups.sh restore samarbeid.claudia-maicher.de

Storage box config

  • Read-only user: ******
  • Read-Write user: ****** 
# samarbeid@dawn:~$ cat .ssh/config
# Host samarbeid-storagebox
#   User ******
#   HostName ******.your-storagebox.de
#   Port 23
#   IdentityFile /home/samarbeid/.ssh/id_ed25519
###

Crontab

user samarbeid on dawn

12 0 * * * cd /home/samarbeid/apps/samarbeid.****** && output=$(./backups.sh create)  && curl -fsS -m 10 --retry 5 -o /dev/null --data-raw "$output" https://hc-ping.com/******
17 0 * * * cd /home/samarbeid/apps/samarbeid-****** && output=$(./backups.sh create)  && curl -fsS -m 10 --retry 5 -o /dev/null --data-raw "$output" https://hc-ping.com/******

Health Checks

Cron + 30m grace time, Pushover + Mail an Benjamin

Edited by Michael Prilop

Merge request reports