device names seen by Veritas. can we rename them?
Created: 31 Oct 2012 | 7 comments
Hello,
I set up Veritas cluster on two systems and both systems are naming same devices differently. see below. powerpath software shows device names same on both systems.
[root@SYSTEM1 ~]# vxdisk -o alldgs list
DEVICE TYPE DISK GROUP STATUS
cciss/c0d0 auto:none - - online invalid
cciss/c0d1 auto:none - - online invalid
emcpowera auto:cdsdisk datadg01 datadg online
emcpowerb auto:none - - online invalid
emcpowerc auto:none - - online invalid
emcpowerd auto:cdsdisk - (syslogdg01) online
emcpowere auto:cdsdisk - (syslogdg01) online
emcpowerf auto:cdsdisk - (syslogdg01) online
[root@SYSTEM2 dev]# vxdisk -o alldgs list
DEVICE TYPE DISK GROUP STATUS
cciss/c0d0 auto:none - - online invalid
cciss/c0d1 auto:none - - online invalid
emc_clariion0_12 auto:none - - online invalid
emc_clariion0_14 auto:none - - online invalid
emc_clariion0_15 auto:cdsdisk - (datadg) online
emc_clariion0_16 auto:cdsdisk - (syslogdg01) online
emc_clariion0_17 auto:cdsdisk - (syslogdg01) online
emc_clariion0_18 auto:cdsdisk - (syslogdg01) online
Is it possible to rename device names which are seen by veritas software when run "vxdisk -o alldgs list" command?
Discussion Filed Under:
Comments 7 Comments • Jump to latest comment
You could try regenerating the names - see extract from Vxvm admin guide:
Mike
UK Symantec Consultant in VCS, GCO, SF, VVR, VxAT on Solaris, AIX, HP-ux, Linux & Windows
If this post has helped you, please vote or mark as solution
I find it strange that Powerpath device names are shown on the one system and enclosure names on the second. Is the same version of Powerpath installed on both systems?
Are the same settings selected for Enclosure Base Naming on both systems?
Even when both systems are using same naming convensions, device numbers are normally different on the nodes - emc_clariion0_12 on system2 may very well be emc_clariion0_1 on system1.
It all depends on the order in which the OS/HBA scans the devices.
Supporting Storage Foundation and VCS on Unix and Windows as well as NetBackup on Unix and Windows
Handy NBU Links
To check the same naming scheme is used on both systems use:
UK Symantec Consultant in VCS, GCO, SF, VVR, VxAT on Solaris, AIX, HP-ux, Linux & Windows
If this post has helped you, please vote or mark as solution
So there are two aspects on the disk names: Disk Media and Disk Acess
Disk media names are those manually assigned when the disk is added to a DG or when the DG is initially create. If the LUNs are shared on both hosts, then the Disk Media names will be the same. These DM names can be changed with vxedit like other objects in the DG
The Disk Access name is what is displayed on the left column of vxdisk list. Its either usually Enclsoure based or OS name. The most recent versions allow these to be changed using utilities that Mike mentioned.
As Marianne mentioned, PowerPath seems to be in use on one node. Is this the also used on the other node ?
cheers
tony
Powerpath was installed on both servers to begin with, but I seem to remember needing to re-install it on server2 for some reason, before I got a successful veritas cluster install. I am showing output from both systems for enclosure setting etc.
It looks to me that server 2 is behaving ok as the disks are named <EnclosueName>_<IndexNumber> as they should be, but server 1 should have disk names something like pp_emc_clariion0_12.
So it MAY work if you rename enclosures so they are the same and regenerate the disk names (using commands as outlined in my first post).
Mike
UK Symantec Consultant in VCS, GCO, SF, VVR, VxAT on Solaris, AIX, HP-ux, Linux & Windows
If this post has helped you, please vote or mark as solution
Are the clariion disks on server2 being managed by Powerpath??
# powermt display dev=all
# powermt display options ### Check if CLARiiON is managed or unmanaged
The fact that server2 is showing emc_clariion0_xx device names rather than pp_emc_clariion_0_xx suggests that VxDMP is being used for multipathing on server2 - this could be bad if Powerpath is also trying to manage the multipathing at the same time!
Check:
server2# vxdisk list emc_clariion0_xx ### pick a disk
compare to server1
server1# vxdisk list pp_emc_clariion0_xx ### pick a disk
Look under Multipathing information (at the bottom)
On server1, would expect to see output as follows:
numpaths: 1
emcpowerXXc state=enabled
On server2, you would see individual ctd paths (2 paths according to your vxdmpadm output)
numpaths: 2
c1tXdXs2 state=xxxxx
c2tXdXs2 state=xxxxx
You mentioned you had to reinstall powerpath on server2 - this is probably the cause of this issue - ie: when powerpath was uninstalled, VxVM picked up the disks as non-powerdevices, so it is trying to manage the multipathing instead of powerpath.
If Powerpath is also managing this device ( powermt display dev=all list the above devices), this is problem.
If server2 is meant to be using Powerpath to manage the devices, you need to do the following:
1. Ensure nothing is using the clariion disks on server2
2. If using earlier version than SF5.1SP1, disable event source daemon
# vxddladm stop eventsource
(see this technote for reference/background: http://www.symantec.com/business/support/index?pag... )
3. For each clariion disk, do # vxdisk rm emc_clariion0_xx
4. Check the disks are being managed in Powerpath correctly, and you have power devices (consult EMC Powerpath documentation for this part as exact commands may depend on your version)
5. Run # vxdisk scandisks
If the powerdevices have been set up correctly, VxVM should pick the devices up as powerdevices now, so vxdisk list should show pp_emc_clariion0_xx disks; and if you run vxdisk list <disk> it should show 1 path to the powerdevice now
6. If you disabled vxesd in step 2, you may want to reenable. See the technote for more details / to determine if this is appropriate for your environment.
If this post has helped you, please vote or mark as solution
Would you like to reply?
Login or Register to post your comment.