Insurance angle is smart. Most people skip that.
OVH Strasbourg fire aftermath, still affecting supply
blujit21 said:
Family members' fiber connections
Must be nice. My aunt's "fiber" in Manila is 35 down, 5 up, and drops when it rains. I tried the distributed home lab thing. Gave up. Now I just rent a cheap Hetzner box and pray.
Rsync over this line takes longer than the backup window.
127.0.0.1 is where the heart is
wendy
Member
frankfurt_ops said:
Please to keep this Documentation with your Offsite rotation logs.
Solid list, though I'd add: know your restore time, not just restore point. We tested our Storage Replica failover last quarter. Took 47 minutes. CEO thought it was "instant." I now put the number in the slide deck in 24pt font.
KnownHost Amsterdam has been stable for us, but I keep Azure as tertiary for the file services. Belt and suspenders.
Have you tried restarting it?
DmitriEng said:
Fire bad for business
"Fire bad for business" is going on my wall next to "temporary atmospheric conversion."
RackNerd's fine for what it is but their ticket responses read like Mad Libs. "Dear [CUSTOMER], we have [VERB] your [NOUN]." Still, beats explaining to a ski lodge why their entire winter is now a 404.
reported. resolved. repeat.
DmitriEng
Member
adam20 said:
Mad Libs
Yes! RackNerd support very robot. But price is real. $18 for year, I pay more for coffee in Moscow. Network to Russia better than OVHcloud was -- https://www.ovhcloud.com
Friend in Poland have 500/500 fiber. Rsync every 6 hour now. I test restore last week, work good. Before fire I never test. Stupid.
This 3-2-1 thing gets repeated like gospel but the real killer is recovery testing under pressure. Anyone can rsync to a second box. Try doing it at 3am when your primary is actually down and your boss is in the Slack channel typing in all caps.
Did the OVH Strasbourg fire take down their public cloud too or just dedicated?
linaplex said:
My aunt's "fiber" in Manila is 35 down, 5 up, and drops when it rains.
That's why I automate the failover, not just the backup. Proxmox HA with corosync, fenced properly. If my sister's power goes out in Ohio, traffic routes to my cousin in Texas within the health check interval. Is it overengineered? Absolutely. Has it saved my ass? Also absolutely.
Cloudflare handles the edge. The origin can literally be on fire and the cached static site stays up.
guru32 said:
Did the OVH Strasbourg fire take down their public cloud too or just dedicated?
Public cloud in that region was affected, yeah. SBG1-4 all had issues. Their status page was a work of fiction for the first six hours though. "Incident under investigation" while photos of the building on fire were already on Twitter.
I was looking at OVHcloud for a project. Went with Hetzner Helsinki instead. Cold, cheap, unlikely to burn.
Zurich1984
Member
jar_lesh said:
Unlikely to burn
Finland has its own risks. I price IPv4 blocks out of Helsinki now and the latency to central Europe is real. Fine for backup, painful for primary if your users are in CH or DE.
Also: OVHcloud's IP pricing post-fire went up. Not dramatically, but enough to notice. Supply constraint in SBG pushed demand to GRA and LIM. Basic economics.
/24 for sale. No lowballs.
sarahjey
Member
As someone who runs everything off a $5 Hetzner box and prays, this thread is terrifying and educational. I have backups to GitHub and that's it. Should I be doing the family fiber thing? I don't have family with fiber.
Is rsync to a friend's box actually reliable enough?
learning on $5 VPS and prayers