I've got an HPE ProLiant ML350 Gen10 with RAID5 configured across five EG001800JWJNL drives running Windows Server 2019 Standard. Last Saturday, one drive failed without any prior alert, and I ordered a replacement that should arrive tomorrow. Then, the next morning, I got a predictive fail alert for a second drive. The server started to slow down, likely due to parity restriping. I had planned to live migrate my Hyper-V VMs to a temporary server, but the building lost power for over an hour before that could happen. Now, the server is stuck in a reboot loop, and I can't seem to stop Windows from attempting to restart after a boot failure. To top it off, my last successful cloud backup was from Saturday morning. I'm working on restoring four VMs from the cloud, but I'm worried that I've lost two days' worth of work and am stuck with no productivity. Does anyone have any tricks or advice?
5 Answers
Sadly, it sounds like you're in a tough spot. Losing two drives in RAID5 isn't good. It's a harsh lesson!
Honestly, for your setup, you should have: 1. RAID1 for your OS, 2. RAID10 for VM data, 3. daily backups, and 4. a cluster to avoid downtime when stuff like this happens. Consider how much it's costing your company to be down—might be time to invest in a more reliable infrastructure.
It's costing us about $18K per day.
This situation highlights why it's best to have the OS and data on different RAID arrays. RAID1 is solid for the OS, while RAID5 or RAID6 is better for data. If you need speed, RAID10 is the way to go, especially for VMs.
Yeah, RAID5 has its issues, especially with virtual machines—RAID10 is preferable.
You might want to check your tape backups if you have them. Good luck!
I'm on that with my cloud backup now.
No magic fix here. Usually, people end up in RAID failure situations at least once in their careers. If you're using larger drives, RAID6 is more forgiving. Remember, the cheaper RAID levels like 5 and 6 aren't favored much anymore—backups are crucial!
If it's for hosting VMs, I agree a RAID10 setup with backups is what I should have.
That's what I was afraid of. Thanks.