# Fallbacks and backups

As a hurricane is barreling towards our machine room in Memphis we are checking our fallbacks and backups for GeneNetwork. For years we have been making backups on Amazon - both S3 and a running virtual machine. The latter was expensive, so I replaced it with a bare metal server which earns itself (if it hadn't been down for months, but that is a different story).

## Tags

* enhancement
* deploy
* assigned: pjotrp

## Tasks

* [ ] Get backups running again on fallback

## Backup and restore

We are using borg for backing up data. Borg is excellent at deduplication and compression of data and is pretty fast too. Incremental copies work with rsync - so that is fast. To restore the full MariaDB database from a local borg repo takes a few minutes:

```
wrk@epysode:/export/restore_tux01$ time borg extract -v /export2/backup/tux01/borg-tux01::BORG-TUX01-MARIADB-20210829-04:20-Sun
real    17m32.498s
user    8m49.877s
sys     4m25.934s
```

This all contrasts heavily with restoring 300GB from Amazon S3.

Next restore the GN2 home dir

```
root@epysode:/# borg extract export2/backup/tux01/borg-genenetwork::TUX01_BORG_GN2_HOME-20210830-04:00-Mon
```

## Get backups running on fallback

Recently epysode was reinstated after hardware failure. I took the opportunity to reinstall the machine. The backups are described in the repo (genenetwork org members have access)

=> https://github.com/genenetwork/gn-services/blob/master/services/backups.org

As epysode was one of the main sheepdog messaging servers I need to reinstate:

* [X] scripts for sheepdog
* [X] enable trim
* [X] reinstate monitoring web services
* [t] reinstate daily backup from penguin2
* [X] CRON
* [X] make sure messaging works through redis
* [x] fix and propagate GN1 backup
* [ ] fix and propagate IPFS and gitea backups

Tux01 is backup up now. Need to make sure it propagates to

* [ ] P2
* [ ] epysode
* [ ] rabbit