Skip to main content

ISC Learn Platform

Restore on Hasdrubal​

Restoring a backup from NAS to Hasdrubal : this is with a separate servername but changing DNS CNAME learn.isc-vs.ch to hasdrubal.isc-vs.ch

Restore on Rumba​

Done and verified 2026-08-01: a full working mirror runs as VM 107 srv-learn-test on rumba — http://learn-test.calypso/learn (VPN only), restored read-only from hannibal with mail muzzled. The complete tested procedure, timings and traps: fromscratch.md. For a real DR cutover the remaining step would be publishing it through srv-web01 under a public name and re-enabling mail + cron — deliberately not done on the standing mirror.

The old idea ("separate servername, change DNS CNAME learn.isc-vs.ch to rumba.isc-vs.ch") still applies for a public cutover, with Caddy on srv-web01 terminating TLS.

Restore inplace on Hannibal​

Restore inplace with the DATA snapshot features​

Restoring the system snapshot : only for a system issue, it wont fix any applicative issue on the Moodle files

** EDIT : After repeated failed tests on Hasdrubal, the system snapshot can have serious issues, for now it is not reliable and can leave the VM with an unbootable system disk. **

Restoring the data snapshot : if the system is down, this can't be used, but on a scheduled Change, we can take a data snapshot to keep a backup of the /srv filesystem (Takes 2h30) and restore Moodle files on a new separate disk (1h on /dev/sdc : this is the way Infomaniak handles filesystem backups, we can't just rollback the current /dev/sdb disk...)

For DRP : only to use in last resort when more recent backup would be corrupted : data snapshots are generally more than weeks old when not taken just before a Change

Restore on a second virtualhost​

Restoring a backup from NAS to Hannibal : This is with a separate servername but changing DNS CNAME learn.isc-vs.ch to the new servername, then switch to it

Restore inplace on the current website​

Restoring a backup from NAS to Hannibal : Directly replacing corrupted files either from the SSD NAS on Rumba, or from the Desktop NAS.

Restore backups from scratch on a new server​

What to do on a new server to get ISC Learn up and running from scratch with only a filesystem backup

Fallback Process​

Steps to get back to normal operation on Hannibal​

Preparation​

  • Schedule a time window to fallback

On D-day​

  • DB backup from the secondary
  • Data backup from the secondary
  • Transfer and restore data files to the primary Hannibal original location
  • Restore DB dump
  • Finish DB restoration from GUI / plugins self checks
  • Clear caches and restart php8.3-fpm / apache2
  • Test and validate everything is back to normal