2013-10-15 01:24 AM
I set up my first real-life cDOT setup yesterday, and I am planning on reusing some diskshelves that still housed some 7-mode aggregates. After changing the ownership of the disks, I can see some remnants of the old, 7-mode aggregates on those disks (from the "disk" command), but I cannot simply offline/destroy those 7-mode aggregates like I was used to doing in 7-mode:
na-croupt::*> disk show
Disk Size Shelf Bay Type Position Aggregate Owner
---------------- ---------- ----- --- ----------- ---------- --------- --------
na-croupt-01:0c.16 827.7GB 1 0 aggregate data aggr0_sata1t
na-croupt-01:0c.17 827.7GB 1 1 aggregate data aggr1_sata1t
na-croupt-01:0c.20 827.7GB 1 4 aggregate data aggr1_sata1t
na-croupt-01:0c.23 827.7GB 1 7 aggregate data aggr1_sata1t
. . .
na-croupt::*> aggr show
Aggregate Size Available Used% State #Vols Nodes RAID Status
--------- -------- --------- ----- ------- ------ ---------------- ------------
239.0GB 11.14GB 95% online 1 na-croupt-01 raid_dp,
239.0GB 11.14GB 95% online 1 na-croupt-02 raid_dp,
2 entries were displayed.
na-croupt::*> aggr delete -aggregate aggr1_sata1t
Warning: Are you sure you want to destroy aggregate "aggr1_sata1t"?
Error: command failed: Aggregate "aggr1_sata1t" does not exist.
I also tried the following, to no avail:
- running the aggr destroy command from the node, but the "destroy" subcommand has been removed.
- halting one node and booting into maintenance mode to destroy the aggregate from there, but I just get:
*> aggr status
the system appears to have no disks!
unable to run aggr command
No root aggregate or root traditional volume found.
You must specify a root aggregate or traditional volume with
"aggr options <name> root" before rebooting the system.
Presumably because a halt of one node moved all the disks to the other node (behavior seems different from 7-mode) (this is a switchless two-node cluster).
My next course of action will be to unassign the disks in cmode, then boot one node into maintenance mode, assign the disks and hopefully destroy the aggregate from maintenance mode, but ideally I would like to know how to solve this problem without having to reboot at all, as I assume this is a scenario that can happen in real-life (i.e. reuse some diskshelves from a 7-mode configuration).
Anybody any ideas ? Thanks in advance!
2013-10-15 03:08 AM
Call support and they can "reenable" something in the nodeshell to delete the aggregate with the node up and running. There is a alternate method to add the aggr to the vldb so the cluster can destroy the aggr, but support will most likely reenable the aggr destroy command from the node temporarily to resolve this. Will take only a few minutes then you can zero and reuse the drives.
Sent from my iPhone 5
2013-10-15 06:03 AM
Very good. I didn't want to post the commands since support should be involved in these. But once you know you can do it again I recommend clearing out on the 7-Mode system before connecting to cDOT but often we may have a shelf and no 7-Mode system to connect first... but a quarantine single node cluster is possible now too... really like where we are at now with cDOT and am using 8.2P4 now on all our installs.