Recently I ran into an issue with Microsoft Cluster Servers (MSCS) on two VMware VMs. The first VM would boot up just fine, but the second VM would take a long time to boot up and sometimes it wouldn’t boot at all. I tracked down the naa of the RDM that was used for MSCS VM and I took a look at the settings of that LUN:

# esxcli nmp device list -d  naa.6090a088006f910fe8df241e1a01d012
naa.6090a088006f910fe8df241e1a01d012
Device Display Name: EQLOGIC iSCSI Disk (naa.6090a088006f910fe8df241e1a01d012)
Storage Array Type: VMW_SATP_EQL
Storage Array Type Device Config: SATP VMW_SATP_EQL does not support device configuration.
Path Selection Policy: DELL_PSP_EQL_ROUTED
Path Selection Policy Device Config: PSP DELL_PSP_EQL_ROUTED does not support device configuration.
Working Paths: vmhba33:C7:T12:L0, vmhba33:C6:T12:L0, vmhba33:C4:T12:L0, vmhba33:C3:T12:L0, vmhba33:C5:T12:L0, vmhba33:C0:T12:L0

# esxcfg-mpath -b -d naa.6090a088006f910fe8df241e1a01d012
naa.6090a088006f910fe8df241e1a01d012 : EQLOGIC iSCSI Disk (naa.6090a088006f910fe8df241e1a01d012)
vmhba33:C7:T12:L0 LUN:0 state:active iscsi Adapter: iqn.1998-01.com.vmware:xxxxx  Target: IQN=iqn.2001-05.com.equallogic:xxxxx Alias= Session=00023d060002 PortalTag=1
vmhba33:C6:T12:L0 LUN:0 state:active iscsi Adapter: iqn.1998-01.com.vmware:xxxxxx  Target: IQN=iqn.2001-05.com.equallogic:xxxxx Alias= Session=00023d050003 PortalTag=1
vmhba33:C4:T12:L0 LUN:0 state:active iscsi Adapter: iqn.1998-01.com.vmware:xxxxx  Target: IQN=iqn.2001-05.com.equallogic:xxxxx Alias= Session=00023d030002 PortalTag=1
vmhba33:C3:T12:L0 LUN:0 state:active iscsi Adapter: iqn.1998-01.com.vmware:xxxxx Target: IQN=iqn.2001-05.com.equallogic:xxxxx Alias= Session=00023d020003 PortalTag=1
vmhba33:C5:T12:L0 LUN:0 state:active iscsi Adapter: iqn.1998-01.com.vmware:xxxx  Target: IQN=iqn.2001-05.com.equallogic:xxxxx Alias= Session=00023d040003 PortalTag=1
vmhba33:C0:T12:L0 LUN:0 state:active iscsi Adapter: iqn.1998-01.com.vmware:xxxxx  Target: IQN=iqn.2001-05.com.equallogic:xxxxx Alias= Session=00023d000002 PortalTag=1

As you can see the SATP used is VMW_SATP_EQL and the PSP used is DELL_PSP_EQL_ROUTED for that LUN. The PSP is DELL_PSP_EQL_ROUTED works in a very similar way as VMW_PSP_RR. All paths are seen as active and IO can go across all the paths. Per KB EqualLogic PS6000X) we can see that a FIXED pathing policy can be used for this Array, so let’s set the pathing policy to FIXED for the RDMs used for MSCS:

# esxcli nmp device setpolicy --psp VMW_PSP_FIXED -d naa.6090a088006f910fe8df241e1a01d012

After that we don’t see all the paths set as working paths:

# esxcli nmp device list -d  naa.6090a088006f910fe8df241e1a01d012
naa.6090a088006f910fe8df241e1a01d012
Device Display Name: EQLOGIC iSCSI Disk (naa.6090a088006f910fe8df241e1a01d012)
Storage Array Type: VMW_SATP_EQL
Storage Array Type Device Config: SATP VMW_SATP_EQL does not support device configuration.
Path Selection Policy: VMW_PSP_FIXED
Path Selection Policy Device Config: {preferred=vmhba33:C7:T12:L0;current=vmhba33:C7:T12:L0}
Working Paths: vmhba33:C7:T12:L0

After setting changing the PSP, both VMs booted up just fine and didn’t have any issue sharing the RDM. Since we are using iSCSI, I would also recommend not presenting RDMs but instead using an in-guest iSCSI initiator to get to a supported configuration. See this KB for more information on supported MSCS configurations.


Published by Karim Elatov

16 April 2012

Tags