EMC PowerPath how to change psuedo names using powermt ?
I hope someone can guide me in the right direction.
I am setting up an nfs failover node.
The NFS servers share out mounts to a SAN.
My problem is this: the master node has
emcpowera as the 100 gig partition
emcpowerb as the 200 gig partition
emcpowerc as the 300 gig partition
and the failover node has
emcpowerc as the 100 gig partition
emcpowerb as the 200 gig partition
emcpowera as the 300 gig partition
This is no problem until the failover node takes over the duties of nfs.
The nfs clients that were connected to the share that map to emcpowerc on the master now for some reason map to emcpowera.
I had a look and see that powermt is responsible for the pseudo naming of the devices.
How do I change the pseudo names for the SAN on the failover node to match the emcpower# (a,b,c) on the Primary node??
Any help will be most appreciated.
Notes I wrote myself a year or so ago:
Use "emcpadm renamepseudo -s emcpower[a-z] -t emcpower[a-z]" to reassign
which pseudo device letter is used for a specific LUN.
e.g. You have LUN 102 showing up as emcpowerg but since it is 3rd LUN you
want it to be emcpowerc instead:
emcpadm renamepseudo -s emcpowerg -t emcpowerc
Thank you so much !
That is spot on exactly what i was looking for !
Saving me so much frustration and nervous exploration.
No problem. I'm sure I probably got that off of the EMC PowerLink site but you're right that sometimes finding things there is a tad difficult.
Greetings to anyone,
is "emcpadm renamepseudo" persistent across reboots?
A positive answer would be appreciated ;)
Thank you in advance,
Yes so far as I know. I initially used the command in 2007 and have not had a problem with the pseudo devices renaming themselves on boots since then.
As noted the main reason I did the rename was because I had two servers in an Oracle RAC setup and wanted both to have same pseudo device name for the shared storage.
What CAN change however, is the /dev/sd* devices that make up the pseudo device. That isn't really a problem so long as you make sure to use the pseudo device name rather than the underlying "real" names.
Running powermt display dev=all will show you which /dev/sd* devices are associated with each of your pseudo devices. I've seen these change in some heavy duty testing back in 2006 (2.4 kernel on that test) but it didn't cause any issues since I was using the pseudo device names in all my specifications.
You want to use the pseudo device anyway because otherwise you're not getting redundant paths.
Thank you for the information,
my problem is rather similar to your:
two nodes have a different naming of /dev/emcpower*
so what on the first node is called emcpowerf on the second is called emcpowerc and so on,
since the oracle service point directly the pseudo devices, I have to
make pseudo device names persisten across reboot,
as you said, I don't care about the odrer of /dev/sd*
Thank you again for your help,
PS: IMHO device-mapper-multipath is far better than PowerPath
P.S. If you click the little thumbs up icon on the post you responded to I'll be "officially" thanked.
Very useful command. It helped me so far.
Error disk in use
# vxdisk list
DEVICE TYPE DISK GROUP STATUS
disk_1 auto:none - - online invalid
emcpowera auto:none - - online invalid
emcpowerb auto:none - - online invalid
emcpowerc auto:none - - online invalid
emcpowerd auto:none - - online invalid
sda auto:none - - online invalid
Since the emc disk was online, the above command failed.
<ERROR: can not use a target device that is in use>
Kindly make sure that the disks are offline befored renaming.
|All times are GMT -5. The time now is 04:52 AM.|