MSCS Cluster - Mount disk iSCSI Failed in Node2

Software-based VM-centric and flash-friendly VM storage + free version

Moderators: anton (staff), art (staff), Max (staff), Anatoly (staff)

Post Reply
itop
Posts: 9
Joined: Thu Sep 21, 2006 6:48 am

Thu Nov 02, 2006 5:32 pm

Hi,

We want to create a cluster MSCS SQL Server 2005 with two nodes which reach the same data via discs iSCSI.
But When the server node2 is started, the discs iSCSI are restored automatically but they are not readable.
In the mmc Disk Management, the two discs iSCSI are marked unreadable.

We have a disc iSCSI for the data of SQL Server 2005 with the letter P: and a disc iSCSI for the quorum with the letter Q:.

Here our architecture:

- 1 server NAS with Windows Storage Server 2003 R2 EN x64. On this waiter, we installed StarWind 3.1.4 and we created 2 IBvolume image.
The first image is for the data of SQL Server and the second image is for the quorum.
On this waiter we have 4 disc SAS in RAID 5.

- 2 server SQL Server with Windows Server 2003 R2 Enterprise EN x64 (SQL1 and SQL2), Microsoft iSCSI Initiator v2.02 and Cluster Service MSCS.
On these two servers, we have map the first disc on the letter P (data) and the second disc on the letter Q (quorum).
On the two servers the service of Cluster MSCS is installed and is configured.

Here how we configured the installation of cluster MSCS on the two machines:
- Shutdown SQL2 and connect to SQL1
- configure Microsoft iSCSI Initiator to connect to two disc iSCSI
- Initialize iSCSI Devices as basic disks.
- When Initialize is completed, create and format new NTFS partition and select drive letter.
- Repeat this procedure for each disk.
- Shut down this server (SQL1). When this server is completely shut down, restart the cluster node 2 server (SQL2).
- configure Microsoft iSCSI Initiator to connect to two disc iSCSI
- The two disks that were formatted on the cluster node 1 server should now appear without any drive letter mounted to them.
Change the drive letter to correspond to those that were used when these drives were formatted.
- Shut down the cluster node 2 server (SQL2). When the cluster node 2 server is completely shut down, restart the cluster node 1 server (SQL1).
- Configuring Microsoft Cluster Service (Name of cluster, account, ..). Configure first node on SQL1
- Add new node : Turn on the cluster node 2 server (SQL2).
- Adding new node is ok, but MSCS don't find disk P: (data) and disk Q: (quorum) because is unreadable.

I try to restart SQL2 and SQL1 but it's the same result.

Do you have an idea of the resolution of the problem?

Thank you in advance for your assistance.

Best regards,
Nicolas COLLET.
Val (staff)
Posts: 496
Joined: Tue Jun 29, 2004 8:38 pm

Thu Nov 02, 2006 9:58 pm

Hi Nicolas,

Have you created the iSCSI targets with 'Clustering' flag ON?
Otherwise such targets will allow only one session at a moment.
Best regards,
Valeriy
itop
Posts: 9
Joined: Thu Sep 21, 2006 6:48 am

Fri Nov 03, 2006 8:56 am

Hi Valery,

Yes, Clustering flag is ON.

This is properties of iSCSI devices :

First for the DATA :
Device: IBV0
Image file name: My Computer\D\ISCSI\cluster-sql-data.ibv
Size: 409600 MB
Mode: Growing image
Asynchronous: No
Last session ID: 1
Snapshot creation period: None
Maximum number of stored sessions: None
Clustered: Yes

Second for the quorum :
Device: IBV1
Image file name: My Computer\D\ISCSI\cluster-sql-quorum.ibv
Size: 1024 MB
Mode: Growing image
Asynchronous: No
Last session ID: 1
Snapshot creation period: None
Maximum number of stored sessions: None
Clustered: Yes

Best regards,
Nicolas COLLET.
Val (staff)
Posts: 496
Joined: Tue Jun 29, 2004 8:38 pm

Fri Nov 03, 2006 11:18 am

Hi,

are there any error messages in the log file?

Have you tried using of ImageFile targets as shared clustr disks?
Usually people use this type of targets building clusters.
Best regards,
Valeriy
itop
Posts: 9
Joined: Thu Sep 21, 2006 6:48 am

Fri Nov 03, 2006 5:19 pm

Hi,

I don't have error in log file.

I have error in event log of second server SQL2 when server is started :
Event Type: Warning
Event Source: mpio
Event Category: None
Event ID: 20
Date: 11/3/2006
Time: 2:44:04 PM
User: N/A
Description:
A Path Verification request to a device on \Device\MPIODisk1 that is controlled by Microsoft iSCSI DSM has failed. This may indicate a Path Failure.

I have this error for each device iSCSI.

I try to create an imageFile target in Starwind and connect to SQL1 and after initialize and format this new device, I add this new ressource in cluster. And when after I configure this new device (ImageFile) in second server SQL2, and reboot, I have the same problem.
ImageFile target is unreadable in SQL2 when SQL1 and SQL2 is started.

Is it possible that the source of the problem comes from architecture 64 bit of the whole of the servers ?

Best regards,
Nicolas COLLET.
User avatar
anton (staff)
Site Admin
Posts: 4008
Joined: Fri Jun 18, 2004 12:03 am
Location: British Virgin Islands
Contact:

Fri Nov 03, 2006 8:12 pm

Can you try to disable MPIO for MS iSCSI initiator?
Regards,
Anton Kolomyeytsev

Chief Technology Officer & Chief Architect, StarWind Software

Image
Val (staff)
Posts: 496
Joined: Tue Jun 29, 2004 8:38 pm

Sat Nov 04, 2006 8:39 pm

Nicolas,
I also think that the MPIO module is to blame.
Idon't sure if the MPIO is compatible with MSCS, aspecially with quorum disks.

So disable or uninstall the MS MPIO driver and try it again.

Please let us know whether it helps.
Best regards,
Valeriy
itop
Posts: 9
Joined: Thu Sep 21, 2006 6:48 am

Mon Nov 06, 2006 12:11 pm

Hi,

I uninstall the MS MPIO Driver for each server and reboot all servers with iSCSI Initiator.
I have the same error, disk is unreadable in second server SQL2.

I have the same error in event log :
Event Type: Error
Event Source: MSiSCSI
Event Category: None
Event ID: 103
Date: 06/11/2006
Time: 12:44:52
User: N/A
Description:
Timeout waiting for iSCSI persistently bound volumes. If there are any services or applications that use information stored on these volumes then they may not start or may report errors.

For more information, see Help and Support Center at [ ... ]
Data:
0000: 5c 00 5c 00 3f 00 5c 00 \.\.?.\.
0008: 73 00 74 00 6f 00 72 00 s.t.o.r.
0010: 61 00 67 00 65 00 23 00 a.g.e.#.
0018: 76 00 6f 00 6c 00 75 00 v.o.l.u.
0020: 6d 00 65 00 23 00 31 00 m.e.#.1.
0028: 26 00 33 00 30 00 61 00 &.3.0.a.
0030: 39 00 36 00 35 00 39 00 9.6.5.9.
0038: 38 00 26 00 30 00 26 00 8.&.0.&.
0040: 73 00 69 00 67 00 6e 00 s.i.g.n.
0048: 61 00 74 00 75 00 72 00 a.t.u.r.
0050: 65 00 64 00 61 00 65 00 e.d.a.e.
0058: 36 00 37 00 32 00 37 00 6.7.2.7.
0060: 65 00 6f 00 66 00 66 00 e.o.f.f.
0068: 73 00 65 00 74 00 37 00 s.e.t.7.
0070: 65 00 30 00 30 00 6c 00 e.0.0.l.
0078: 65 00 6e 00 67 00 74 00 e.n.g.t.
0080: 68 00 36 00 33 00 66 00 h.6.3.f.
0088: 66 00 61 00 62 00 37 00 f.a.b.7.
0090: 32 00 30 00 30 00 23 00 2.0.0.#.
0098: 7b 00 35 00 33 00 66 00 {.5.3.f.
00a0: 35 00 36 00 33 00 30 00 5.6.3.0.
00a8: 64 00 2d 00 62 00 36 00 d.-.b.6.
00b0: 62 00 66 00 2d 00 31 00 b.f.-.1.
00b8: 31 00 64 00 30 00 2d 00 1.d.0.-.
00c0: 39 00 34 00 66 00 32 00 9.4.f.2.
00c8: 2d 00 30 00 30 00 61 00 -.0.0.a.
00d0: 30 00 63 00 39 00 31 00 0.c.9.1.
00d8: 65 00 66 00 62 00 38 00 e.f.b.8.

Do you have other idea ?

Thanks in advance.

Best regards,
Nicolas COLLET.
Val (staff)
Posts: 496
Joined: Tue Jun 29, 2004 8:38 pm

Mon Nov 06, 2006 12:31 pm

Nicolas,

It's very strange.
Could you send me the StarWind Service's log file for this case?
Best regards,
Valeriy
Post Reply