Matt, we can't retain vol0 for a couple of reasons -
1) The old heads are x86 (V3020) and new heads are x64 (N6040 / V3120), therefore the binaries are different. 2) NetApp ONTAP on old heads, IBM ONTAP on new ones.
How about if I do a config dump and a config dump -v on the old heads and do a config restore on the new ones before swinging the aggrs?
They are both running 7.2.6.1 (I had to downlevel the n6040 FROM 7.2.3 for snapmirror purposes).
Bob McKim
|------------> | From: | |------------>
----------------------------------------------------------------------------------------------------------------------------------------|
|Matt Hallmark matt@cosmictoaster.com |
----------------------------------------------------------------------------------------------------------------------------------------|
|------------> | To: | |------------>
----------------------------------------------------------------------------------------------------------------------------------------|
|Bob McKim/Omaha/IBM@IBMUS |
----------------------------------------------------------------------------------------------------------------------------------------|
|------------> | Cc: | |------------>
----------------------------------------------------------------------------------------------------------------------------------------|
----------------------------------------------------------------------------------------------------------------------------------------|
|------------> | Date: | |------------>
----------------------------------------------------------------------------------------------------------------------------------------|
|08/05/2010 05:34 PM |
----------------------------------------------------------------------------------------------------------------------------------------|
|------------> | Subject: | |------------>
----------------------------------------------------------------------------------------------------------------------------------------|
|Re: A couple of V-Series Questions |
----------------------------------------------------------------------------------------------------------------------------------------|
For #1, you are creating a new aggr and migrating data. the aggregates on a V-Series are raid0 (aggr options aggrname to see it), so you can't fail a drive/lun per se.
For #2, you are correct on most of the steps, but as you will be retaining your root volumes, there's no need to re-create shares or change dns etc.
Procedurally it's:
1. Halt existing cluster and power off the nodes 2. Rack and cable the new cluster, power on the nodes and go into maintenance mode. 3. Verify that the ports you want to use are initiators and document the WWPN's (storage show adapter) 4. Update the zoning on the switches for the new initiator ports 5. On the back-end SAN, update your host profile to reflect the new WWPN's 6. From maint mode, reassign the LUNs from the old systemID to the new one, repeat for the other node 7. using aggr status from maint mode, validate that you can see your various aggrs and that you have one marked as a root aggr on each controller 8. halt both nodes and boot them up normally. 9. Upgrade ontap if you are changing CPU architectures.
As a pre-step, you might want to validate that the version of OnTap that's on the CF of the new controllers is compatible w/ the version of ontap (raid labels etc). If the existing OS/raid labels etc in-use is 7.2.X, you will very likely want to upgrade to 7.3.X before doing the head swap, as the last V-Series I un-boxed last month was running 7.3.2GA.
Matt Hallmark
On Aug 5, 2010, at 2:44 PM, Bob McKim wrote:
- I have a customer who accidentally added a LUN to an existing aggr
that was incorrectly sized. What is the correct procedure to replace that LUN with a correctly sized one? They can't resize the LUN on the existing array as it's out of space. Do I create a correct sized LUN on another array, add it as a spare and then do a disk fail against the wrong sized LUN? I've never tried this on a gateway.
- Does anyone have some procedures for a headswap on a v-series/gfiler?
There appears to be a "black hole" on the NOW site in regards to how-to upgrade a v-series/gfiler.
Is it as simple as -
- Gather existing v-series head configuration
- Zone and LUN mask the new v-series head to the existing LUNs
- Perform a disk reassign ownership on an aggr by aggr basis
- Create new CIFS shares on new v-series head (only CIFS shares on
existing v-series head)
- Remove old v-series head from DNS (or swap servers in DFS)
- Add new v-series head to DNS
- Verify client connectivity
Thanks
Bob McKim
Although no hardware was injured in the creation of this note, a few
billion electrons and photons were utilized against their will.
