Import fails: One or more devices are missing from the system.

[titan_rw]

I suspect the missing device was stripped into the pool as a regular (non redundant) vdev, and not as a slog.

Normally that kind of zdb output is indented for ease of reading, but if I'm reading it correctly, your pool has 6 vdevs:

Code:

Configuration for import:
  vdev_children: 6

Children 0, 1, 2, 3, are complete, with two disks per vdev.

Children 5 is your slog, which is complete, and not a degraded mirror:

Code:

[*]children[5]:
[*]  type: 'disk'
[*]  id: 5
[*]  guid: 13566463784014113451
[*]  whole_disk: 1
[*]  metaslab_array: 47
[*]  metaslab_shift: 31
[*]  ashift: 12
[*]  asize: 399995043840
[*]  is_log: 1

Children 4 is what I presume is your missing disk. It doesn't indicate it was ever a slog, but a separate single disk vdev:

Code:

[*]children[4]:
[*]  type: 'missing'
[*]  id: 4
[*]  guid: 0

If you no longer have this missing disk, the pool is lost. The issue was that this disk was added into the pool as a single non redundant data disk, not as a mirror component to a slog vdev.

 

[Rand]

No,
the only thing i could think of was a missing second log device (mirror'd), but that was the only thing I (thought I might have) changed.

Edit Re missing vdev.
Hm, now that you mention that, i had that before once, but I am not sure whether it was this particular pool which suffered from this.
Have too many things on my mind and too many places testing and fixing stuff at the same time.

Since this sounds like the most logical explanation (given the evidence) you might be right.

Thanks a lot fall all the advice.
Copying the backup files back has been completed this morning, so nothing lost but lessons learned:)