Saturday, August 18, 2012

Adding a slice to Solaris 10 from EMC storage

Hi All, I have assigned a 100 GB on raid 1 from my EMC to 2 "clustered" servers & when running the format command Iam able to view the disk added, but how do i make it usable for my server & being a clustered server what are the things I should keep in mind. My disk is shown as 17. c5t600601606CC12400244CB4AC733EDF11d0 /scsi_vhci/disk@g600601606cc12400244cb4ac733edf11 I have run the format command & selected the disknumber 17 & format was completed successfully.

SUN cluster software is being used & the LUN i have to assign is from an external storage box(EMC CX4 120c) Fc disk. I already have luns assigned from the same box to these servers. The new LUN is required for moving the Archive logs. We dont have any agent running to the server by executing the format command Iam able to see the all the LUN's including the new LUN I have assigned from EMC. How do I make it usable now?
Iamable to see the slice on my Solaris server. now i want to format the disk & mount the disk on my production server
 What cluster software are you running ? Sun Cluster or Veritas Cluster ? also, what multipath software are you using? vxdmp or mpxio ? I see that maybe is mpxio installed. if so please paste the output of mpathadm list lu 

I use a sun cluster, I dont have any multipath or agents running on mu server for connecting the EMC storage.

it appears you have emcpower devices listed in your format command so 
you are using EMC's Powerpath for mulitpathing. You should check with
your storage team to see which device name you should be using to access
the disk.
hi there 
if u r usinf EMC powerpath ,use these commands in u r solaris to see path 
#powermt check 
#powermt config 
in here u will see the path & the slice so that u can use ur slice for 
hope this works
Is this filesystem goint to be part of Sun Cluster? If so, you have to add it to the actual cluster configuration ( resource group ). 
If not, you should treat it as a normal filesystem/lun, but sun clsuter must be aware of that: 

The steps to add the filesystem in only one node ( local filesystems, not HA filesystems ) should be like this: 

1) discover the lun in the node 
2) issue sgdevs ( to create the new disk ids) 
3) tell to sun cluster that this disk will be treat it as local disk: 
scconf -c -D name=dsk/d[N],localonly=true 

4) work with your local filesystem. 

If its not a local filesystem, you must to create a resource and tell it which resource group will contain it: 
1) discover the lun in both nodes 
2) do your multipath stuff. 

3) Add the new filesystem into the /etc/vfstab. Sun cluster takes the information from there when you create storage resources for the cluster. 
Remember we always in sun cluster works with Disk Devices ( aka sun cluster dids ), not disk paths (no SVM paths, nor powerpath Paths, nor mpxio paths) 

The path that you must to use for LUNS is the output of the did devices in cldev list - cldev show ( and find then your lun/mapping corresponding to the disk id that you must enter into the vfstab ). If you are not sure about this, check the documentation about "disk devices or sun cluster dids". Otherwise, the failover for that mountpoint could fail. 
Do this question to you: What if in node A the disk is /dev/dsk/c4t0d0s0 and in node B the disk is /dev/dsk/c5t0d0s0? 
How sun cluster will know about this disk name change? That is one of the reason because sun cluster created the "dids" (disk ids) 

4) if you don't have a resource group, then create it: 

a) clrg create rerouceGroupName 

Add the resource to rerouceGroupName 
b)clrs create -g rerouceGroupName -t SUNW.HAStoragePlus:4 -p FilesystemMountPoints=/mountPoint resource-name 

c) if the resource is not enabled, then enable it: 
clrs enable resource-name 

check the status: clrs status resource-name 

d) Important step: perform a failover =) 


SUNW.HAStoragePlus:4 is the version for your HAStoragePlus Resource agent. You can see your agents vesions with: 
a) clrt list 

Check your sun Cluster documentation at sunsolve. The above steps worked for me lot of times, there is no warranty about this explanation. 


Post a Comment

Design by BABU | Dedicated to grandfather | welcome to BABU-UNIX-FORUM