OK, W set to zero and uploaded tracks with
no errors. I guess that means that your modified hijack is what
fixed the problem.
Great. I'll make it an "official" release shortly.
Just out of interest, will enabling/disabling
write-caching have any performance benefits/detriments
in this environment?
I think I'd suggest disabling it, as our kernel doesn't know
anything about safely flushing the write cache, and
(contrary to common opinion) the empeg s/w *does*
write to the media
(dynamic data partition)during normal playback operation.
So just leave write-caching
off for safety with these slow-writing SSDs.
For regular hard disks, no issues either way.
Do you have any idea why I was unable to use /dev/hdb
to specify the slave SSD with hdparm?
No. But that *has to work* if you want a two-drive system.
I'm probably just missing something obvious..
Cheers