1. Will the current LUN be acessible to the host during the Raid Group expansion?
Yes, the existing LUN will be restriped across the disks in the RAID Group.
2. Can I use diskpart to expand the LUN on the host, using concatonation to allow the host immediate access to the increased size.
I believe you will have to wait until the restriping is complete before the additional space is available. There used to be an option to expand the LUN with the RAID Group if the existing LUN used occupied the entire RAID Group but the option is not available if you have metaLUNs.
So you will need to create a new LUN and concatenate it to the existing LUN.
2. Can I use diskpart to expand the LUN on the host, using concatonation to allow the host immediate access to the increased size.
Host Info: 2 Windows Servers 2003 in a clustered environment.
yes you can ..there are a few caveats you need to know dealing with expanding storage on ms clusters ...Aran is our cluster-master ...he can elaborate more on that
I would think that the clustered disk resource would have to be online as I believe if you take the clustered disk resource offline you will loose access to it and therefore cannot expand it with diskpart...?
I would furthermore then do a fail-over to the second node to confirm that it accepts the new disk size but no expansion should be required.
As in all cases a backup prior to commencing the diskpart procedure would be highly recommended.
In MSCS environment, if you are not running a volume manager like Veritas Storage Foundation, you are using basic disks and you can use diskpart to expand the LUN: 1) Increase the LUN on the CX3-80 2) Expand the partition with diskpart on the active node 3) No expansion to be done on the passive node
Are you sure ? I remember having read somewhere that the cluster resource must be taken offline as well as the standby node. Then enlarge the LUN and bring everything up again.
I have made the operation several times without taking offline the LUN. But if it is written somewhere to it is recommended, I would like to know where and why.
Everybody recommends backing up first: MS as well as EMC. About having the cluster resource taken offline, I'm not sure. But expanding on 1 node is enough, the other node will work just fine when the other node did the expansion.
Are you sure ? I remember having read somewhere that the cluster resource must be taken offline as well as the standby node. Then enlarge the LUN and bring everything up again.
yep ..like calle said ..if you take disk resource offline in cluster, you will not be able to see it in Disk Management ..and diskpart will not be able to expand it either. The only issue i've see is where diskpart does its job on active node but when your fail the resource over to the standby node ..it does not see the new space until after it's rebooted.
it does not see the new space until after it's rebooted
That's what I meant. I think we meant the same thing but didn't say it right. So the 2nd node needs a reboot anyway: in that case you can shut it down (node = down, so offline), do the expansion and bring the 2nd node up again. We meant exactly the same thing.
You can perform the expansion of a disk assigned to a cluster on the active node with the passive node(s) online (sorry I am late to the conversation dynamox, pacific time and all that). The disk will show as expanded on the passive nodes(s) after it is expanded on the active node, but the disk resource will not failover successfully until the passive node(s) have been rebooted.
I read somewhere that you need to put the 2nd node offline, but cannot fint the document anymore. It might have been a Microsoft document... I don't remember.
That might have been for an older rev of the OS? I have tested this in Windows Server 2003 clusters under i/o load and it worked very well.
I would like to say THANKS to everyone who answered. Like I said I've just added a single disk to a raid group for expansion, usually did a seperate radi set then a metalun. I guess it sort of the same idea just a bit of a different twist.
AranH1
2.2K Posts
0
July 7th, 2008 14:00
during the Raid Group expansion?
Yes, the existing LUN will be restriped across the disks in the RAID Group.
using concatonation to allow the host immediate
access to the increased size.
I believe you will have to wait until the restriping is complete before the additional space is available. There used to be an option to expand the LUN with the RAID Group if the existing LUN used occupied the entire RAID Group but the option is not available if you have metaLUNs.
So you will need to create a new LUN and concatenate it to the existing LUN.
dynamox
9 Legend
•
20.4K Posts
0
July 7th, 2008 16:00
using concatonation to allow the host immediate
access to the increased size.
Host Info: 2 Windows Servers 2003 in a clustered
environment.
yes you can ..there are a few caveats you need to know dealing with expanding storage on ms clusters ...Aran is our cluster-master ...he can elaborate more on that
calle2
133 Posts
1
July 8th, 2008 01:00
I would furthermore then do a fail-over to the second node to confirm that it accepts the new disk size but no expansion should be required.
As in all cases a backup prior to commencing the diskpart procedure would be highly recommended.
ovivier
142 Posts
0
July 8th, 2008 01:00
In MSCS environment, if you are not running a volume manager like Veritas Storage Foundation, you are using basic disks and you can use diskpart to expand the LUN:
1) Increase the LUN on the CX3-80
2) Expand the partition with diskpart on the active node
3) No expansion to be done on the passive node
RRR
4 Operator
•
5.7K Posts
0
July 8th, 2008 01:00
RRR
4 Operator
•
5.7K Posts
0
July 8th, 2008 02:00
In a cluster I can imagine that if the 2nd node is still seeing the LUN while it's being expanded by the 1st node he could get confused....
ovivier
142 Posts
0
July 8th, 2008 02:00
RRR
4 Operator
•
5.7K Posts
0
July 8th, 2008 02:00
RRR
4 Operator
•
5.7K Posts
0
July 8th, 2008 02:00
About having the cluster resource taken offline, I'm not sure. But expanding on 1 node is enough, the other node will work just fine when the other node did the expansion.
dynamox
9 Legend
•
20.4K Posts
0
July 8th, 2008 04:00
the cluster resource must be taken offline as well as
the standby node. Then enlarge the LUN and bring
everything up again.
yep ..like calle said ..if you take disk resource offline in cluster, you will not be able to see it in Disk Management ..and diskpart will not be able to expand it either. The only issue i've see is where diskpart does its job on active node but when your fail the resource over to the standby node ..it does not see the new space until after it's rebooted.
RRR
4 Operator
•
5.7K Posts
0
July 8th, 2008 05:00
That's what I meant. I think we meant the same thing but didn't say it right. So the 2nd node needs a reboot anyway: in that case you can shut it down (node = down, so offline), do the expansion and bring the 2nd node up again. We meant exactly the same thing.
AranH1
2.2K Posts
0
July 8th, 2008 07:00
AranH1
2.2K Posts
0
July 8th, 2008 07:00
offline, but cannot fint the document anymore. It
might have been a Microsoft document... I don't
remember.
That might have been for an older rev of the OS? I have tested this in Windows Server 2003 clusters under i/o load and it worked very well.
dynamox
9 Legend
•
20.4K Posts
0
July 8th, 2008 07:00
SanMan5
25 Posts
0
July 8th, 2008 07:00
Thanks Again,
Bing