CDP Filter driver deploy fails

[JaySt]

i'm trying out CDP. I'm deploying it using the latest v11 install (25th may patch included) against a vSphere 7 environment.
In my source vCenter, i have 2 clusters. The first cluster for which i tried to deploy the IO filter went smoothly. No issues.
The second cluster went less smooth while ESXi versions are the same, network config are the same etc.
When i deployed the filter driver on the second cluster, the deployment seems to go OK, but after the first run of the deployment wizard, there was a notice the IO filter needed to be upgraded.
I was presented with the option to upgrade the driver, but that process fails due to a maintenance mode requirement. I thought this was very strange compared to the same process on the first cluster. Uninstalling the driver is not working either due to the same maintenance mode requirement. Maintenance mode is a bit difficult at the moment, so we'll have to postpone.


both clusters are running ESXi 7.0.1.16850804 (7.0 update 1)

I'm planning on creating a Veeam support case, but perhaps someone has any insights on where the difference in behaviour between the two clusters comes from regarding filter deployment.

[HannesK]

Hello,
yes, the case number would be interesting. As you noticed, that's unexpected behavior that should be checked and fixed if necessary.

Thanks,
Hannes

[wishr]

Hi JaySt,

It's hard to tell why exactly the maintenance mode is required - it can only be concluded after analyzing vCenter and host logs.

However, I should clarify a few things.

First, all the requirements about maintenance mode and reboots come directly from the vSphere environment. Basically, we use API calls to deploy the filter, and if for whatever reason vSphere identifies that a maintenance mode is required to accomplish that, you see a corresponding message in the VBR UI.

Next, we have not observed situations when putting hosts into maintenance mode is required during the initial I/O filter installation. Usually, it is only required when you are performing an I/O filter update or uninstalling it, so I suspect there was a previous I/O filter version installed on that cluster or it has not been uninstalled properly.

Anyway, the best bet would be to contact our tech. support representatives so they would assist you with investigating and resolving the issue.

Thanks

[JaySt]

case 04840991

[JaySt]

Support confirmed we need to look at the vSphere hosts to discover the reason why these specific hosts are not able to install the filter driver without going into maintenance mode first. There has to be an underlying cause on the ESXi host level. I'll see if i can get some logs with clues.

[JaySt]

did some more troubleshooting.
the cluster with filter deployment issues required maintenance mode to proceed. the hosts at that time -did- have a veeamcdp filter registered under configure->storage->I/O Filters within vcenter host configuration. We first had to reboot the host. After the reboot, the filter was not present anymore. Veeam VBR was not aware of this fact. Refresh did not sync this state, VBR still thought it had something on the hosts. Only option was to uninstall the filter driver and see what would happen. The uninstall process failed obiously, but did reset state to the correct one (no filter installed). After that, VBR would let us install the filter driver succesfully.

i'll leave it at that.

[HannesK]

thanks for posting the result of the case