So my "automated backup" from Hetzner turned out to be a snapshot of a running system lol I go to restore after a kernel panic and the mysql tables are just gone corrupted mid-snapshot I guess support said "restore from your own backup" which I didnt have because I thought their backup WAS my backup so now im testing everything Who else actually tested their provider restore? Be honest
How do you test without breaking the live site? I have a small shop on RackNerd and never dared click the restore button. Scared it will overwrite something
🚫 = mas tipo eu fiz restore test no CloudCone kkkk Hostinger backup corrompido mano
Member
---
- restore testing:
- method: spin up parallel VPS
- note: NOT in production account
- warning: some providers link accounts by payment
- my results:
- Contabo: snapshot booted, data stale by 6h
- KnownHost: "backup" was live rsync, incomplete
- Leaseweb: actual working restore
- NOTE: Leaseweb took 4h to provision restore target
- WARNING: GreenCloudVPS restore deleted source VM during test
---
Member
Actually your Method is flawed oder? Live snapshot corruption is known Problem. Actually providers should use Filesystem quiescing or Application consistent hooks. Actually most dont. I tested Twelve providers. Four passed. Hetzner failed. OVHcloud failed. HostHatch failed. Vultr passed. Actually verify Checksums oder?
Restore testing is not optional 😊 I test monthly my provider failed last week now I run 3-2-1 myself no provider trust anymore
