toadstool331
asked on
VSphere 5 iSCSI Datastore not found
Question: Why did I lose my datastore?
The environment is (2) esxi hosts (5.0.1, and 5.1) with iSCSI (Synology software iSCSI)
In the process of shutting down and moving hardware (no changes) , the iSCSI datastore was not found after startup by either host. iSCSI LUN was found and all paths checked out, however it was unable to mount datastore.
The only way I was able to mount datastore was to re-signature datastore with new UUID. This datastore is only used for backup, and we were able to make necessary changes. Just need to understand what happened.
The environment is (2) esxi hosts (5.0.1, and 5.1) with iSCSI (Synology software iSCSI)
In the process of shutting down and moving hardware (no changes) , the iSCSI datastore was not found after startup by either host. iSCSI LUN was found and all paths checked out, however it was unable to mount datastore.
The only way I was able to mount datastore was to re-signature datastore with new UUID. This datastore is only used for backup, and we were able to make necessary changes. Just need to understand what happened.
ASKER
Synology model is DS1512+ (listed in compatibility guide)
Yes, the LUN was presented. When adding datastore, first option to mount existing (without modifying sig) failed repeatedly.
Yes, the LUN was presented. When adding datastore, first option to mount existing (without modifying sig) failed repeatedly.
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ASKER
What is the easiest way after the re-signature for the guest OSs to connect to the updated datastore?
Guest OS, this was used as an RDM?
Guest VMs are normally located on the datastore?
what's the issue?
Guest VMs are normally located on the datastore?
what's the issue?
ASKER
The guest OS cannot start because it does not see the hard disk vmdk files on the iscsi datastore. After the re-signature of the datastore the datastore name has been changed. There is something that needs to be edited I believe to tell the guest where to find those vmdk files now.
Check the datastore is not missing the disks?
Easiest way, would be to remove the virtual disks from the VM inventory, e.g. remove from VM, do not delete from disk.
Add the disks again, by browsing the datastore.
Does this issue, of datastore loss, happen alot?
you may want to check your ESXi/ESX logs (/var/log/vmkernel.log)
but the logs are no presistant between reboots on ESXi, unless you are logging to persistent storage of Remote Syslog Server.
and look for path loss, information
Easiest way, would be to remove the virtual disks from the VM inventory, e.g. remove from VM, do not delete from disk.
Add the disks again, by browsing the datastore.
Does this issue, of datastore loss, happen alot?
you may want to check your ESXi/ESX logs (/var/log/vmkernel.log)
but the logs are no presistant between reboots on ESXi, unless you are logging to persistent storage of Remote Syslog Server.
and look for path loss, information
ASKER
Trying to remove the virtual disk from the vm results in an error, Invalid configuration for device '0'. It will not remove the virtual disk.
Unregister the VM from the inventory. (do not delete from disk).
edit the VMX file with VI, WinSCP, and Remove the reference to the virtual disks.
make sure you have a backup of the VMX file.
edit the VMX file with VI, WinSCP, and Remove the reference to the virtual disks.
make sure you have a backup of the VMX file.
ASKER
Thanks Savant!
No problems, you also missed MVE, & Expert of the Year 2013, 2011, 2012!
keep an observation on your logs!
keep an observation on your logs!
I had a similar problem with a HA cluster, where the passive server was lost during firmware upgrades.
Problem:
On two ESXi 5.1 hosts, I was able to remount a volume using the vSphere Client after selecting the corresponding path and choosing "use existing signature" for mounting.
On a third ESXi 5.1 host, I only was offered the option "format", while "use existing signature" and "resignature" as well where in-active.
Cause: It appears that the volume was mistaken for a snapshot by this host.
Solution:
I logged into the esxi-host by ssh and checked for snapshots by
# esxcli storage vmfs snapshot list
Subsequently I mounted the volume by
# esxcli storage vmfs snapshot mount -l <label>|-u uuid
I was surprised but glad this worked even though the vSphere GUI refused to do such.
More details can be found in KB article 1011387 from vmware, e.g. you may need to unmount the "snapshot" as such after mounting the volume.
Problem:
On two ESXi 5.1 hosts, I was able to remount a volume using the vSphere Client after selecting the corresponding path and choosing "use existing signature" for mounting.
On a third ESXi 5.1 host, I only was offered the option "format", while "use existing signature" and "resignature" as well where in-active.
Cause: It appears that the volume was mistaken for a snapshot by this host.
Solution:
I logged into the esxi-host by ssh and checked for snapshots by
# esxcli storage vmfs snapshot list
Subsequently I mounted the volume by
# esxcli storage vmfs snapshot mount -l <label>|-u uuid
I was surprised but glad this worked even though the vSphere GUI refused to do such.
More details can be found in KB article 1011387 from vmware, e.g. you may need to unmount the "snapshot" as such after mounting the volume.
You probably found the LUN was presented, to the Hosts, but the datastore was missing?
You are not the first to ask this question on EE, it's asked about 30 times a year!
Is your Synology iSCSI SAN on the VMware HCL ?