I'm sorry but I've moved away from Amanda so can't help. However one thing that your description does imply is that you keep the drives in the enclosure next to the machine. Unless you have some other mechanism for taking a remote backup, this is a bad idea. consider the aftermath of a fire, theft, electrical malfunction or virus infection. Adjacent drives may have either disappeared, melted, had their interfaces ruined or been corrupted. In any case, quite possible no use as backups. Best practice is to ensure that periodically a good backup (equivalent of a level 0) is taken and stored offsite. For domestic users that might be stored in a safe elsewhere, but offsite is the ideal. Sorry to nag, but better to act now than when you are faced with a new insurance replacement and no useable backup. Martin. On 18/09/2021 08:05, Jon LaBadie wrote: > My backup system (using amanda) stores its data on external > drives in a single 4 bay USB enclosure. As these drives > are only needed during backup or recovery operations I have > used hdparm to enable them go to an idle state after a period > of inactivity. > > If I attempt to access any of the data when the drives are > sleeping I get a strange result. Suppose I try to list > one of the mountpoints, "ls .../D2", there is the expected > delay while the 4 drives spin up and then ls completes with > no output. > > I know each of the 4 drives has 40 subdirectories under the > mount point. And each of the 3 other reawakened drive lists > properly. But the directory I used to awaken the drives > lists as empty. > > This effect is not limited to inititally listing a mount point. > Had my command been "ls .../D2/DS1-044", DS1-044 would appear > empty, but DS1-043 and all other similar directories list > properly. > > Further, if I attempt to access a file I know exists in DS1-044 > by its explicit name, that succeeds. It is like having execute > permission, but not read permission on the directory. > > If I unmount and re-mount the filesystem, all is normal. > > Any clues as to why this happens, or ways to make the invisible > visable again without the unmount/mount sequence? > > Jon > -- J Martin Rushton MBCS