Repository navigation
Read-write SSD cache on third-party M.2 NVMe — working procedure for the "failed to mount an SSD read-write cache" error #605
Replies: 2 comments
|
If you scheduled syno_hdd_db.sh with the The |
|
Thanks — that explains it perfectly. I've bumped my pinned version to v3.6.137 (which also picked up the E10M20-T1 db-file fix from today) and switched my scheduled runs to |
Uh oh!
There was an error while loading. Please reload this page.
After whitelisting with syno_hdd_db.sh (thanks for this script!), DSM still refused to create a read-write cache on my third-party NVMe (2× Samsung PM983 on an E10M20-T1, DS3622xs+, DSM 7.4): "The system failed to mount an SSD read-write cache". Root cause: DSM re-locks unassigned third-party M.2 to read-only during the cache-creation flow's NVMe rescan, so sfdisk fails with Operation not permitted even when the drives show compatibility=support.
The fix is holding the devices writable during creation — once created, the cache persists across reboots with no further hacks. I documented the full procedure (plus step ordering, a verify script, and Task Scheduler persistence) here: https://github.com/illofspeed/synology-ds3622xs-unsupported-drives
Also confirms: never use --force — it's what forces the M.2 read-only in the first place. Validated across reboots and a full NVMe swap (Micron 7450 → Samsung PM983).
All reactions