The question is whether we can learn something from these experiences.
As far as I can see our SCSI error handling code is independent of
the devices concerned.
But I am very surprised when a disk times out during normal operation,
especially if there have not been any medium errors just before.
On the other hand, there are somewhat more `suspect' devices.
Worst of all are scanners. But also CDROM and tape are orders
of magnitude less reliable than disks.
Are there reasonable error recovery schemes where the collection of
outstanding requests and the devices involved are taken into account?
And are there reasonable error recovery schemes that work from the
bottom instead of the top?
[I mean: right now we have the famous ( abort / bus device reset /
bus reset / host reset ) sequence. One might also start at the
other end and upon a timeout ask (i) is there still communication
with the host adapter? (ii) are other devices such as the disk
still functioning? Etc.]
Andries
-
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to [EMAIL PROTECTED]
