Hello everyone,
Due to high electricity prices, I'm starting to experiment with pushing my linux NAS into ACPI
S3 at night.
the embedded system has no other option than using "echo mem > /sys/power/state", or rtcwake.
Unfortunately, when waking up (ether-wake or rtcwake), the MD driver starts kicking out SCSI
disks from a RAID6 array.
SAS HBA initialization is slow due to a large nr of disks, so some are not yet available at that
time.
when you see in dmesg, a few seconds later, they appear.
Is there a way to tell the md raid subsystem "wait xx seconds after resuming from S3 before you
decide to kick out disks"?
another option would be: the journal on the RAID volumes is a 60GB SSD partition,
is there an option to kick out preventively and re-add when it appears, simply by replaying the
journal?
Thank you,
Janpieter Sollie
