Subscribe
Accepted Solution

Snapdrive for UNIX 5.0 : export bug ?

Hi,

I have a problem with Snapdrive for Unix 5.0. When I start a "snap connect" SDU 5.0 doesn't correctly create NFS export.

Here is my configuration : a Netapp with 2 interfaces, one for management (10.0.38.10) and one for data (192.168.0.10). Everything is properly configured in SDU :

[root@lab-oracle ~]# snapdrive config list

username            appliance name   appliance type

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

root                3020demo1        StorageSystem

[root@lab-oracle ~]# snapdrive config list -mgmtpath

system name   management interface   datapath interface

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

3020demo1     10.0.38.10             192.168.0.10

Here is the problem

1/ Create a snapshot

[root@lab-oracle log]# snapdrive snap create -fs /home/oracle/app/oracle/oradata/orcl/data -snapname test

Starting snap create /home/oracle/app/oracle/oradata/orcl/data

  WARNING:  DO NOT CONTROL-C!

            If snap create is interrupted, incomplete snapdrive

         generated data may remain on the filer volume(s)

         which may interfere with other snap operations.

Successfully created snapshot test on 192.168.0.10:/vol/LABORACLE_DB

        snapshot test contains:

        file system: /home/oracle/app/oracle/oradata/orcl/data

        filer directory: 192.168.0.10:/vol/LABORACLE_DB

2/ Mount the snapshot

[root@lab-oracle log]# snapdrive snap connect -fs /home/oracle/app/oracle/oradata/orcl/data /mnt/test -snapname  192.168.0.10:/vol/LABORACLE_DB:test

connecting /mnt/test

          to filer directory: 192.168.0.10:/vol/LABORACLE_DB_0

        Volume copy 192.168.0.10:/vol/LABORACLE_DB_0 ... created

                 (original: LABORACLE_DB)

Successfully connected to snapshot 192.168.0.10:/vol/LABORACLE_DB:test

        file system: /mnt/test

        filer directory: 192.168.0.10:/vol/LABORACLE_DB_0

3/ Snapdrive 4.2 properly create NFS export with root=192.168.0.57:10.0.38.57, not SDU 5.0 !

SDU 4.2

/vol/LABORACLE_DB_0     -sec=sys,rw,root=192.168.0.57:10.0.38.57

SDU 5.0

/vol/LABORACLE_DB_0     -sec=sys,rw

This causes problems with SMO jobs especially during verification or clone because SMO has no right to write on the clone

server.log.8:2012-02-20 16:47:25,677 [btpool0-19 - /smo_v9/services/SMO] [DEBUG]: OperationCycle status: FAILED, rootErrorCode: 13032, rootErrorMessage: SMO-13032: Cannot perform operation: Backup Create.  Root cause: SMO-11007: Error cloning from Snapshot copy: FLOW-11019: Failure in ExecuteConnectionSteps: SD-10038: File system /opt/NetApp/smo/mnt/-home-oracle-app-oracle-oradata-orcl-ctrl2-20120220164625801_0 is not writable.  Please ensure that the SnapManager process has write access to the file system.  After correcting this, you may need to take another snapshot., opId: ff808081359b707401359b707db20001

Any idea ? Bug or not ?

Re: Snapdrive for UNIX 5.0 : export bug ?

Got news from Netapp support : this SDU 5.0 bug will be fixed soon

Re: Snapdrive for UNIX 5.0 : export bug ?

All,

this has been reported in BURT 57541: http://support.netapp.com/NOW/cgi-bin/bol?Type=Detail&Display=575741&app=portal

Title :

SnapDrive 5.0 for UNIX operations may fail in NFS environments as SnapDrive prevents a clone volume from inheriting the exports option of the parent volume

Workaround : Modify the export option of the parent volume and replace b root=<hostname/IP>b  with b anon=0b  and then retry the SnapDrive or SnapManager operation.

Notes : The above defect will be fixed in the P1 patch of SnapDrive 5.0 for UNIX