LTO3 - move to EOD failed / transport error

wbk

Renowned Member
Oct 27, 2019
204
26
68
Hi all,

I have a container backed on a mediaset of five LTO3 tapes (tape0-tape4).

The fifth tape now returns an error. I am aware that LTO3 is not officially supported, but in practice things use to turn out quite well. This is the log of the task:

Code:
2025-05-31T22:34:14+02:00: update media online status
2025-05-31T22:34:14+02:00: media set uuid: 8e8a53fb-7b10-4ca1-a224-b57b451d59b7
2025-05-31T22:34:15+02:00: found 8 groups (out of 8 total)
2025-05-31T22:34:15+02:00: backup snapshot "ct/104/2024-03-30T17:26:36Z"
2025-05-31T22:34:19+02:00: allocated new writable media 'osbas4'
2025-05-31T22:34:19+02:00: Checking for media 'osbas4' in drive 'delltape'
2025-05-31T22:35:00+02:00: found media label osbas4 (bb818850-3efa-4a31-b3d1-cbabb53259c9)
2025-05-31T22:35:00+02:00: moving to end of media
2025-05-31T22:45:32+02:00: queued notification (id=ab234de4-0cbd-467e-bf1f-817dc1906f99)
2025-05-31T22:45:32+02:00: TASK ERROR: move to EOD failed - scsi command failed: transport error

I have two external tape drives, one LTO3 and, more recently, one LTO4; both of them have a SAS connection. Before that I ran an external LTO3 drive over old fashioned SCSI, but I am quite sure this mediaset never saw that drive (it is not in my list of devices in PBS, or I have replaced it with the newer SAS device).

The thing is, I recall starting to use tape4 on the SAS LTO3 device. I am not sure whether after replacing the device with the LTO4 one, I was ever able to make a recording, but I am sure that when I was not able to do so, I blamed the drive in the first place.

Now that it really was time to get back ups running, I replaced the LTO4 device with the LTO3 one, which in turn required a cleaning cycle for the first time since I got it. Without a cleaning tape, it took some time to get hold of one. Now that it has had its cleaning cycle, I started the backup of my datastore/container.

The device took its time moving to the end of the media before throwing the tape back out as seen in the task log.

In short:
  • 5 tapes in the mediaset (all LTO3)
    • tape0 - full
    • tape1 - full
    • tape2 - full
    • tape3 - full
    • tape4 - writable
  • 3 drives
    • SAS LTO4 (maybe once used succesfully with tape4, maybe only errors)
    • SAS LTO3 (tapes 0-3, probably 0-4)
    • SCSI LTO3 (prodably not involved in this mediaset, though it may be that the first tapes in the set are written with that drive)

What I want to ask is:
  • Can using the LTO3 tape in an LTO4 drive make the tape unusable in an LTO3 drive?
    • I supposed it can mix and match, because of backwards R/W compatibility a single generation
  • Should I just wipe the mediaset, format tapes and start over?
    • I still have multiple backups of this container in the datastore
    • I would lose historic snapshots
  • Is there an overview of which backup is spread over which tapes of the mediaset?
    • In case data on tape4 is lost/inaccessible, can I find out whether any of the backups is restorable without trying each of them?
    • Because of deduplication, I can imagine data being spread over all of the tapes
  • Since I am not quite sure about the change from SCSI LTO to SAS LTO3, in combination with the direct access to the drive from PBS and possible low level changes between SCSI and SAS, and the non-supported state of these generations, I can imagine this playing part as well.
Since I still have multiple backups of the container in the datastore, the risk of either outcome is low. I mostly want to prevent building up incorrect habits or expectations of the possibilities.