November 8, 2008 at 5:25 pm
Hi All,
We have SQL Server 2005 cluster setup as follows:
3 node active/active/passive cluster with 1 instance on node1 and 5 instances on node3.node is passive.
Active(node1) Passive(node2) Active(node3)
INS1 INS2
INS3
INS4
INS5
INS6
Node1 has drives Node3 has drives
C-> OSINS2-> G->primary
D->Primary H->log ]Group2
E->secondary T->tempdb
F->log]group1
Q->quorum
S->tempdb INS3-> I->primary
Y->msdtc J->log ]group3
Z->backup U->tempdb
INS4-> K->primary
L->log ]group4
V->tempdb
INS5-> M->primary
N->log ]group5
w->tempdb
INS6-> O->primary
p->log ]group6
R->tempdb
Group1,2,3,4,5,6 are groups in cluster administrators.
Here are the Issues:
1.The drives of node1 are also showing up in node3?(happend after doing some upgrade in SAN)->major issue
2.If want to install more instances like instance7 we have no drive letters to assign?
3.How can we install 26 failover instances if we have this drive letter conflict?
plz advice me
Thanks
Madhu
November 9, 2008 at 7:42 am
Hi Madhu,
In answer to your first question; SQL Server 2005 supports mounted volumes so you can overcome this drive letter shortage, you can read about it here:
"SQL Server 2000 and SQL Server 2005 support for mounted volumes ", http://support.microsoft.com/kb/819546. (it also contains a link to "How to configure volume mount points on a Microsoft Cluster Server").
Can you access the disks fron Node 3 as well? If I were you I would go back to the SAN vendour and point this out. You definitley don't want to have any question marks around disk arbitration/reservation in your database cluster.
HTH!
/Elisabeth
elisabeth@sqlserverland.com
MCITP | MCT
http://sqlblog.com/blogs/elisabeth_redei/
http://linkedin.com/in/elisabethredei
November 9, 2008 at 12:23 pm
thank you Elisabeth,
The drives of node are also showing up in node3 along with node3 drives.
and when I try to access the node1 drives from node3, its giviming the error access denied.
what would be the possible reasons for showing up drives of node1 in node3? We also have a developement cluster setup, some of the development cluster drives are also showing up in node3 of production cluster?
any help would be appreciated.
Thanks
Madhu
November 10, 2008 at 6:18 pm
Hi,
anybody come across this type of drives inconsistency in cluster nodes?
could you plz shed some light on this..
Thanks
November 11, 2008 at 4:01 am
Hi,
If this is a SCSI cluster then it is not that unusual to see this behaviour and I have had this myself (with a certain version of the drivers involved).
I it's a fibre/SAN storage there is something flaky with the storage driver; as I said before, check with the guys who did the upgrade.
/Elisabeth
elisabeth@sqlserverland.com
MCITP | MCT
http://sqlblog.com/blogs/elisabeth_redei/
http://linkedin.com/in/elisabethredei
November 11, 2008 at 9:05 am
Mandhu,
Is this the same infrastructure / servers we were talking about the other day? If so I dont really get where the 7 instances came from.
If your Node 1 drives appear on your Node 3 servers then your disk/volume presentation and permissioning is messed up.
If Node 1 can see Node 3's drives but not access that’s how it should be (the drives should appear but be inaccessible until the cluster is failed over).
N.B: :exclamationmark: This is pretty advanced stuff, we were talking the other day about a small cluster setup and I mentioned that you must be clear what you are trying to achieve and know all about SAN and Windows clustering. What your talking about here is a lot more complex and I really feel (bearing in mind it sounds like your volume assignment and presentation structure is floored) that you need to seek professional advise. You could quite easily lose a LUN or re-initialise a volume and lose a lot of data! :exclamationmark:
Adam Zacks-------------------------------------------Be Nice, Or Leave
November 11, 2008 at 1:23 pm
Thanks
Actually we have 1 instance on Node1(Active) and 4 on Node3(Active) ,Node2 is Passive.
I mentioned 7instances here to clarify the Lack of drive letters issue, if we want create more instances.
Ok..In general,If Node1(Active) has D,E,F drives and Node3(Active) has G,H,I
Then
1.D,E,F of Node1 can visible in Node3 and are not accessible? and
2.G,H,I of Node3 can visible in Node1 and are not accessible? is this normal behaviour?
3.In this case what are the drives visible in node2?(passive)
In my case the drives are showing like the below:
Node1 Node2 Node3
(Active) (Passive) (Active)
C C
G C
D H D
E I E
F J F
J K G
S L H
U M I
W N J
X T K
Y U L
Z V M
W N
X Q
Y S
Z T
U
V
W
X
Y
Z
November 11, 2008 at 10:14 pm
It's the storage controller not letting go of the drives. Please contact the people who did the upgrade and inform them. Only they will know if this is anything to worry about with this particular storage controller and the set of drivers it is using.
/Elisabeth
elisabeth@sqlserverland.com
MCITP | MCT
http://sqlblog.com/blogs/elisabeth_redei/
http://linkedin.com/in/elisabethredei
November 12, 2008 at 2:49 am
madhu.arda (11/11/2008)
Thanks1.D,E,F of Node1 can visible in Node3 and are not accessible? CORRECT. BUT THEY SHOULD NOT SHOW
2.G,H,I of Node3 can visible in Node1 and are not accessible? CORRECT. BUT THEY SHOULD NOT SHOW
3.In this case what are the drives visible in node2?(passive) D, E, F, G, H, I BUT NOT ACCESSIBLE.
Basically live nodes should not be able to see drives from other live nodes (if they are not a passive failover node for that server).
In your case as you have explained Node 1 should not be able to see Node 3 drives. Node 3 should not be able to see Node 1 drives and Node 2 (passive) should be able to see all drives from Node 1 and Node 3 but not access them until a particular nodes shared services/drives are failed over This assumes Node 2 is the passive fail over node for Node 1 and Node 3).
Adam Zacks-------------------------------------------Be Nice, Or Leave
Viewing 9 posts - 1 through 8 (of 8 total)
You must be logged in to reply to this topic. Login to reply