We're getting the error below on an EqualLogic PS4000e SAN array for each of the three volumes every couple of minutes. I've searched for this issue and they recommended making sure the option "Consult locally defined CHAP accounts first" is on (it is). Snapshots are enabled on all three volumes but are not scheduled. We also have had two instances in the last month where the host server spontaneously disconnected the three logical drives connected to this array, which we've not been able to isolate a cause for, but suspect might be related. A reboot of the host server got everything connected back up.
Level: ERRORTime: 12/28/12 3:12:14 PMMember: WEMG01Subsystem: MgmtExecEvent ID: 7.4.3iSCSI login to target '192.168.1.4:3260, iqn.2001-05.com.equallogic:0-8a0906-0883c390a-a8a000000164deff-workingdata' from initiator '192.168.1.6:60725, iqn.1991-05.com.microsoft:server.domain.com' failed for the following reason: Initiator cannot access this target because an iSCSI session from another initiator already exists and multihost access is not enabled for this target.
Any ideas folks?
Thanks in advance!
Are you running a clustered file system against that volume? Based on the volume name, I suspect not. To prevent accidental damage from having multiple servers connecting to one volume you have to explicitly enable multihost access. Otherwise only the first server to log on will be allowed to login.
So filesystems like MS Clusters, Hyper-V clusters, ESX VMFS, tivoli fs, etc... do allow multiple hosts to access one volume without damage.
If you are just using standard NTFS, EXT3/4 filesystems then DON'T enable Multihost access!!! Other corruption WILL result.
You don't want to allow multiple hosts to even access a volume if a cluster filesystem isn't in place. So please change your volume ACLs so that only one host can even see a volume.
I.e. its very dangerous to have asterisks in your IP based ACLs or a single CHAP username for multiple servers.
Someone later on can access those volumes and reformat them. I have had to rebuild many VMFS partitions (not always successfully) because a Windows server attached to a VMFS volume and reformatted while the ESX servers were still attached.
The three volumes have NTFS file systems and are connected to a single host machine. The host does have three Hyper-V VMs on it, and the SAN volumes store the VHD files, but the VMs don't connect directly to the array.
We didn't set up this system, so I'm not sure if it was done correctly or not.
Thanks in advance,
If this isn't a 2008R2 Hyper-V cluster then only one host should have access to a volume.
If you check the volume ACLs you may find that a Wildcard or common CHAP account is in use.