Hello kubaq87,

Which JBOD’s do you have as you didn’t state them in your post? I am guessing that you have a MD1000,12xx, or 14xx enclosure.  If that is the case & your virtual disk is spanning both enclosures, then you can’t remove one enclosure without harming your virtual disk.  All your virtual disk information is stored on your PERC card & on the drives that make up the virtual disk.  If you move the enclosure, then you will need the same perc card type that it is connect to see the virtual disk on the new host.   

Please let us know if you have any other questions.

Hello Sam,

Thanks for your reply. Maybe I did not state this correctly, but every disk is in it's own DG. So in other words there is not Drive Groups between enclosures. This is how does the summary looks like:

# ./SMcli 172.16.75.103 -c "show storageArray summary;"
Performing syntax check...
                  
   STORAGE SUMMARY                                            
   Disk pools:                                     0          
      Virtual Disks on Disk Pools:                 0          
   Disk groups:                                    120        
      RAID 0 Disk Groups: 120  Virtual Disks: 120             
   Access virtual disks:                           1          
   Standard Virtual Disks (Used/Allowed):          120 / 512  
      Base:                                        120        
      Repository:                                  0          
   Thin Virtual Disks (Used/Allowed):              0 / 512    

   HOST MAPPINGS SUMMARY                                                  
   Access virtual disk:     LUN 31,31 (see Mappings section for details)  
   Default host OS:         Windows (Host OS index 0)                     
   Mapped virtual disks:    121                                           
   Unmapped virtual disks:  0                                             


   HARDWARE SUMMARY                                                                    
   Enclosures:                                  2                                      
      System configured to use batteries:       Yes                                    
                                                                                       
   RAID Controller Modules:                     2                                      
      Consistency mode:                         Duplex (dual RAID controller modules)  
                                                                                       
   Physical Disks:                              120                                    
      Current physical disk media types:        Physical Disk (120)                    
      Current physical disk interface type(s):  Serial Attached SCSI (SAS) (120)                                
                                                                                                                      
   FEATURES SUMMARY                                              
   Feature enable identifier:  37303030323930303033363554B8374F  
   Feature pack:               SAS - MD3460                      
   Feature pack submodel ID:   215                               

   Additional feature information                                             
      Snapshot groups allowed per source virtual disk (see note below):  4    
      Virtual Disks allowed per storage partition:                       256  


   FIRMWARE INVENTORY                                          
   MD Storage Manager®                                         
      AMW Version:              11.20.0G06.0020                
      Report Date:              Tue Jun 05 08:54:14 CEST 2018  
                                                               
   Storage Array                                               
      Storage Array Name:       med-file-md12                  
      Current Package Version:  08.20.05.60                    
      Current NVSRAM Version:   N2701-820890-004               
      Staged Package Version:   None                           
      Staged NVSRAM Version:    None                           
                                                               
   RAID Controller Modules                                     
      Location:                 Enclosure 0, Slot 0            
      Current Package Version:  08.20.05.60                    
      Current NVSRAM Version:   N2701-820890-004               
      Board ID:                 2701                           
      Sub-Model ID:             215                            
                                                               
      Location:                 Enclosure 0, Slot 1            
      Current Package Version:  08.20.05.60                    
      Current NVSRAM Version:   N2701-820890-004               
      Board ID:                 2701                           
      Sub-Model ID:             215                             
                                                     
   Physical Disk     
      Enclosure, Drawer, Slot:           Manufacturer:     Product ID:          Physical Disk Type:            Capacity:        Physical Disk firmware version:     FPGA Version: (SSD only)  
      Enclosure 0, Drawer 0, Slot 0      SEAGATE           ST4000NM0023         Serial Attached SCSI (SAS)     3,720.523 GB     GS0F                                Not Available              
      Enclosure 1, Drawer 3, Slot 11     SEAGATE           ST4000NM0023         Serial Attached SCSI (SAS)     3,720.523 GB     GS0D                                Not Available             
      Enclosure 1, Drawer 4, Slot 0      SEAGATE           ST4000NM0023         Serial Attached SCSI (SAS)     3,720.523 GB     GS0D                                Not Available             
      Enclosure 1, Drawer 4, Slot 1      SEAGATE           ST4000NM0023         Serial Attached SCSI (SAS)     3,720.523 GB              

Script execution complete.

SMcli completed successfully.

And what I am trying to do is to disconnect bottom enclosure(Enclosure 1), and attach it to different server. As you see they both belong to same array, but there is not disk groups created on between enclosures. In other words each DG has only one physicall member, so all drives are separated(we have RAID on OS level).

120 physicall disks, and 120 separated DGs(I removed most of them from CLUI output, because it was too big)

So the question is, if in this case I can disconnect Enclosure 1 without destroying array ?

Hello kubaq87,

From the output that you posted, you are running an MD3460 & MD3060e.  Based on that no you can’t separate your MD3060e and keep the configuration on those drives.  The reason is that your MD3460 is controlling current configuration, & your MD3060e is present to it.  If you remove your MD3060e and don’t destroy the virtual disks & disk groups you will put both controllers into a lockdown state.  In addition to that when you connect your MD3060e to another host your configuration will not be seen due to the DG & VD configuration being on your MD3460.

Please let us know if you have any other questions.

Hello Sam,

Many thanks for your advices. So just last one question to summarize this. Will this work if I do it this way:

1. Destroy all Virtual Disk/Drive Groups on MD3060e

2. Remove MD3060e from array

3. Attach MD3060e to another server

4. Create new array on MD3060e

Hello kubaq87,

What is the OS that your host is running that your VD are presented to?  Also, what software are you using for your software raid? Can we get a support bundle from your MD3460 so that we can review it?  I can send you can email that you can reply with the support bundle log. It is also a good thing to have a support bundle handy just in case the system has any issues we can see what the config was like and fix any data base issue.

The steps you listed are correct.  I would also get a valid & test backup of the data on your MD’s prior to doing any steps just in case something goes wrong. What it would be best to do is to power down your MD3460 & MD3060e and then disconnect the cables from your MD3060e then power on your MD3460 & make sure that it doesn’t go into a lockdown state, & to make sure that you can still access the data on your MD3460.  If you can still access your data on your MD3460 then you can move your MD3060e to your new host and clear the configuration & create the new virtual disk configuration.

In most cases when you want to remove an expansion enclosure it is best to destroy all DG & VD then recreate as the controller can go into a lockdown state due to the change.   

Please let us know if you have any other questions.

Sorrry, I thought we switched to email, and forgot topic. I sent support bundle via email. Regarding questions: We are running CentOS 7.4 with ZFS pool as a "RAID" solution.