disarraydo you all trust raid 6?
I know raid 10 is better that's not what I am asking
I know raid 10 is better that's not what I am asking
スプリットショックウイルス †@disarray In what way do you mean trust?
@disarray also how large is the array?
Phoenix@disarray yes. a friend of mine's NAS that I use runs on raid 6, with 9 drives. Sustaining two drive failures is good enough for me tbh
@disarray something something always have multiple backupd anyways
wlo@disarray it depends on what's being kept on it amd what the required uptime is.
iirc if it's object storage for some fedi instances, absolutely.
iirc if it's object storage for some fedi instances, absolutely.
LisPi@disarray It's fucked on #btrfs so no.
It's ostensibly okay on #ZFS. But I'm using cheap & slow HDDs so resilvering risk means nope. For an SSD array it's perfectly fine.
More generally I don't really trust #filesystems, maintaining #DataIntegrity is all nice & sweet but without #DataConsistency it loses a lot of practical value.
Shit still gets corrupted unless you're using a #database atop those integrity-preserving filesystems.
@disarray If cost is a motivation for the question, consider that used enterprise SAS (and SATA) drives are cheap-enough you can easily go raid1c3 or raid1c4 without actually paying more than if you bought a single such drive new.
Obviously this requires a disk array or a server with many bays.