I have a 2-node switchless cluster. I would like to shut it down. I disable the cluster. I have disk ownership as shown. once I halt both node and boot them. I found the following result. The controller that does not own the disk container whre its root partition reside has no root volume to boot up. BTW I wonder why disk is reserved and "waiting for giveback" while cluster HA is disabled.
::> disk show -partition-ownership Disk Partition Home Owner Home ID Owner ID -------- --------- ----------------- ----------------- ----------- -----------
Info: This cluster has partitioned disks. To get a complete list of spare disk capacity use "storage aggregate show-spare-disks". 1.0.0 Container NetappClu01-01 NetappClu01-01 538012598 538012598 Root NetappClu01-02 NetappClu01-02 538013735 538013735 Data NetappClu01-01 NetappClu01-01 538012598 538012598
Feb 21 11:13:45 [NetappClu01-01:cf.fmns.skipped.disk:notice]: While releasing the reservations in "Waiting For Giveback" state Failover Monitor Node State(fmns) module skipped the disk 0b.00.3 that is owned by 538012598 and reserved by 538013735. Waiting for reservations to clear Waiting for reservations to clear Feb 21 11:14:59 [NetappClu01-01:sas.link.error:error]: Could not recover link on SAS adapter 0a after 45 seconds. Offlining the adapter. Feb 21 11:16:16 [NetappClu01-01:config.invalid.PortToPort:error]: SAS adapter "0a" is attached to another SAS adapter. Feb 21 11:17:11 [NetappClu01-01:sas.link.error:error]: Could not recover link on SAS adapter 0a after 45 seconds. Offlining the adapter. Waiting for reservations to clear Waiting for reservations to clear
Controller B result:
Feb 21 03:14:13 [localhost:raid.assim.tree.noRootVol:error]: No usable root volume was found! Uptime: 2m8s System rebooting...
I want to conduct power maintenance. I have to shudtdown the 2-node cluster. so I disable the cluster HA by command "cluster ha modify –configured false". Then I halt both controller into loader mode. Then both node cannot boot up. you can see result in the first post.
That’s not the normal steps at all. You never need to disable HA to power off systems (in general you should never disable HA on two node cluster unless following well defined procedure). I suspect OP confused HA and storage failover.
Because storage failover was not inhibited, one node is in failover. Now if it is also node that has epsilon (HA is disabled), this is really a problem. I agree, better open support case, it goes above normal forum level.