Funny disk name

[CGCHEN555]

Hi All,

We added 3 new draws today and we noticed one of the draws had funny device names shown on the partner node. The node itself is displaying the correct names. Contacted support and was told to halt both filers and restart the draw in question before bringing the filers back. I'll like to avoid doing this for obvious reasons.  Does anyone know how we can workaround this issue?

Thanks.

Partner controller:

Partner disks

RAID Disk    Device      HA  SHELF BAY CHAN Pool Type  RPM  Used (MB/blks)    Phys (MB/blks)

---------    ------      ------------- ---- ---- ---- ----- --------------    --------------

partner     3d.SHJ0000000054B0.0     3d	0   0   SA:B   0  SAS  15000 0/0	560879/1148681096
partner     3d.SHJ0000000054B0.2     3d	0   2   SA:B   0  SAS  15000 0/0	560879/1148681096
partner     3d.SHJ0000000054B0.14    3d	0   14  SA:B   0  SAS  15000 0/0	560879/1148681096
partner     3d.SHJ0000000054B0.1     3d	0   1   SA:B   0  SAS  15000 0/0	560879/1148681096
partner     3d.SHJ0000000054B0.5     3d	0   5   SA:B   0  SAS  15000 0/0	560879/1148681096
partner     3d.SHJ0000000054B0.12    3d	0   12  SA:B   0  SAS  15000 0/0	560879/1148681096

Primary controller:

spare	3d.01.0     3d	1   0   SA:A   0  SAS  15000 560000/1146880000 560879/1148681096
spare	3d.01.1     3d	1   1   SA:A   0  SAS  15000 560000/1146880000 560879/1148681096
spare	3d.01.2     3d	1   2   SA:A   0  SAS  15000 560000/1146880000 560879/1148681096
spare	3d.01.3     3d	1   3   SA:A   0  SAS  15000 560000/1146880000 560879/1148681096
spare	3d.01.4     3d	1   4   SA:A   0  SAS  15000 560000/1146880000 560879/1148681096
spare	3d.01.5     3d	1   5   SA:A   0  SAS  15000 560000/1146880000 560879/1148681096
spare	3d.01.6     3d	1   6   SA:A   0  SAS  15000 560000/1146880000 560879/1148681096

[aborzenkov]

Such names are seen in case of duplicate shelf id. Verify that shelf has unique id. And yes, to change SAS shelf id you need poser cycle it.

[CGCHEN555]

We did have duplicate shelf ID and that was fixed and the new draw was power cycled but not the controllers. Any solutions of not having to halt both filers to fix this?

[aborzenkov · Accepted solution]

You could try takeover/giveback (which is in effect rebooting controller). But in general changing shelf configuration is twilight zone where everyone tends to be on the safe side and say it must be done offline.

[CGCHEN555]

Thanks for that, I'll give that a go.

[CGCHEN555]

Thanks for that. fail over fixed the problem