mxraid_assemble: malfunction if enclosures are registered too late

[thomas]

Sometimes RAIDs are not assembled at startup, reason is that mxraid_assemble is called before enclosure/disk initialisation has finished.

Some dmesgs from phaedra:

Enclosure discovers disks ...
[    9.243472] scsi 16:2:0:0: Direct-Access     SEAGATE  ST8000NM0065     K004 PQ: 0 ANSI: 6
               14 more in between, approx. 40 ms apart
[    9.619708] scsi 16:2:15:0: Direct-Access     SEAGATE  ST8000NM0065     K004 PQ: 0 ANSI: 6

     *** TOO EARLY ! ***
mxraid_assemble finished it's job on 2 disks for scratch/local
[    9.655871] md: md0 stopped.
[    9.661035] md0: detected capacity change from 0 to 4000529252352

The lazy enclosure continues ....
[    9.725575] scsi 16:3:121:0: Enclosure         AIC 12G  3U16SAS3swap     0c01 PQ: 0 ANSI: 5
[    9.727575] scsi 16:3:123:0: Enclosure         ADAPTEC  Smart Adapter    4.02 PQ: 0 ANSI: 5
[    9.762259] sd 16:2:0:0: Attached scsi generic sg3 type 0
               again 14 more in between, approx. 0.3 ms apart
[    9.769710] sd 16:2:15:0: Attached scsi generic sg18 type 0

Now devices are assigned (mind the interleave)
[    9.764620] sd 16:2:0:0: [sdd] 1953506646 4096-byte logical blocks: (8.00 TB/7.28 TiB)
[    9.774241] sd 16:2:4:0: [sdh] 1953506646 4096-byte logical blocks: (8.00 TB/7.28 TiB)

So waiting 0.2 secs would have helped in this case.

[thomas]

1f5d973 tries to cope with the situation

[thomas]

#67 should have solved the issue, but keep an eye on it (Especially with phaedra).