Unsolved
This post is more than 5 years old
2 Posts
1
1270
VSM selective restore doesn't work
Environment description:
3x ESXi hosts (Dell-ESXi-5.5U3-3116895-A02) in a cluster, vSphere 5 Essentials Plus license
vCenter Server Appliance (ver. 5.5.0.30500 Build 4180648), vCenter Server 5 Essentials license
Dell VSM 4.5.2.974
2x PS6210 in a group, single pool, over 40 datastores, mostly around 500 GB, some up to 2TB
Issue:
when trying to use selective restore for a VM it always ends up with an error. It's during the task to "Mount datastore for VM restore". The related errors are:
No PS snapshot was brought online for datastore {datastore name}
Rescan did not find added VMFS disk: {disk iqn}
I rerun the restore while looking at PS group manager GUI. The VSM creates a temporary clone from snapshot, but the ACL generated is incorrect. That is likely the root cause. It's worth to mention that the volume doesn't have the "VSMAR-AUTO- policy set up. It has only a manually created access policy group, including all 3 ESXi hosts iSCSI initiators iqns and pair of IP addresses per host.
When the temporary cloned volume is created it is assigned an access policy, but in a wrong way, e.g. the iSCSI initiators IP addresses are mangled, the first figure of the first octet is missing and there is only 1 IP address instead of two per host.
Should the selective restore work even on a volume not created by VSM, but which has VSM scheduled snapshots?
Is there a way to recreate the VSMAR-AUTO- policy in order to fix this ?
Many thanks
jan.rendos
2 Posts
0
December 15th, 2016 05:00
I had a look at the ACL as visible in VSM, and found out that if there is an extended access point within a policy set up in the dell PS group manager with multiple IP addresses they appear in VSM in this format:
[ip1, ip2]
and the field has red borders, which indicates that the format is incorrect.
So I added cluster access to the ACL in VSM and it actually added additional lines, one per ESXi host. Each of them with only iSCSI initiator iqn (no IP addresses).
After this I looked at the ACL in DELL PS group manager and the group ACL was gone and instead there is a new policy called VSMAR- , which includes the newly added lines and mangled records from the original group ACL. By mangled I mean that instead of 2 IP addresses per host there is only one per host (actually the first one), BUT
AND THIS IS IMPORTANT
the first figure of the first octet is missing.
e.g. original for the first host was 10.11.12.13, 10.11.12.14
after VSM did create a new one it changed to 0.11.12.13
Even if the IP would be correct, still the other IP address is missing which might cause the temporary volume not being visible for the ESXi host if that ESXi interface would be down at that moment of rescan.