Timeline for mdadm repair single chunk / sector
Current License: CC BY-SA 4.0
19 events
when toggle format | what | by | license | comment | |
---|---|---|---|---|---|
Aug 17, 2018 at 9:46 | vote | accept | user10186803 | ||
Aug 7, 2018 at 15:03 | answer | added | user10186803 | timeline score: -1 | |
Aug 7, 2018 at 10:17 | history | edited | user10186803 | CC BY-SA 4.0 |
added 17 characters in body
|
Aug 6, 2018 at 15:07 | comment | added | Attie | Also this: jrs-s.net/2015/02/03/will-zfs-and-non-ecc-ram-kill-your-data | |
Aug 6, 2018 at 15:01 | comment | added | Attie | "Thank you for your input, but I disagree" ... fair enough. All I will say is that it sounds like you're planning to run a production storage array with the day-to-day mentality that should only ever be considered when you're in "oh sh*t... now what" recovery mode. | |
Aug 6, 2018 at 14:46 | history | edited | user10186803 | CC BY-SA 4.0 |
Tried to explained the parts of my question that are not duplicate
|
Aug 6, 2018 at 14:38 | answer | added | Deltik | timeline score: 1 | |
Aug 6, 2018 at 14:32 | comment | added | user10186803 | And I don't see how this is a duplicate. My original question is how to mark a sector as bad and force mdadm to repair this sector. I also don't think it's an XY problem since I know all the data and offsets, I only need to tell one program to use the data collected from an other program. Maybe this is not possible in this situation? | |
Aug 6, 2018 at 14:29 | comment | added | user10186803 | @Attie Thank you for your input, but I disagree on the statement that ECC memory for ZFS is just a small bonus and not a requirement. And yes, they do: link | |
Aug 6, 2018 at 14:15 | review | Close votes | |||
Aug 22, 2018 at 3:05 | |||||
Aug 6, 2018 at 14:11 | comment | added | Attie | Possible duplicate of Btrfs over mdadm raid6? | |
Aug 6, 2018 at 13:58 | comment | added | Deltik | Even though the question asked is not the same as Btrfs over mdadm raid6?, this is an XY problem that is fully addressed by the other question and answer. | |
Aug 6, 2018 at 13:56 | comment | added | Attie | "Netgear's ReadyNas and Synology combine mdadm and btrfs, and still keep bitrot protection" Do they? | |
Aug 6, 2018 at 13:53 | comment | added | Attie | You're not just protecting against bit rot, but also other things like read / write errors (e.g: high write). The ZFS / ECC issue has been hugely exaggerated and misunderstood - yes a running machine might benefit from ECC, but for data to suffer a number of rare issues will have to occur in just the right way. You would be better off using ZFS for the situation you've outlined... How would using BTRFS+MDADM+Scripts without ECC be less of an issue than ZFS without ECC? | |
Aug 6, 2018 at 13:36 | comment | added | user10186803 | If I use ZFS without ECC, I could just as well not worry about bitrot protection. Both prevent very rare errors, but I want to do this right. As for being too hands on, you're right. But I don't see any better way. I know it's possible, netgear's ReadyNas and Synology combine mdadm and btrfs, and still keep bitrot protection. | |
Aug 6, 2018 at 13:30 | comment | added | Attie | "... I then can translate this offset to a disk and a offset on that disk ..." - you are planning on being far too hands-on with the storage... It will probably go wrong. | |
Aug 6, 2018 at 13:27 | comment | added | Attie | You do not need ECC memory to use ZFS.... I'd recommend you use ZFS. | |
Aug 6, 2018 at 13:25 | review | First posts | |||
Aug 6, 2018 at 13:41 | |||||
Aug 6, 2018 at 13:20 | history | asked | user10186803 | CC BY-SA 4.0 |