icingacli vspheredb check vm --name <name> fails with non-ascii output

[peteeckel]

Expected Behavior

icingacli vspheredb check vm --name <name>

returns the status of the VM in the vShpereDB module.

Current Behavior

For all but one VM, that is the case.

There is, however, one new VM that refuses to show up in the GUI and returns a binary blob in the CLI after clearing the screen:

#`▒▒▒][q▒x▒6w_O"

Piping the output to less to avoid the screen clearing reveals the output in more readable form:

[UNKNOWN] Failed to load object "K<CA>G^L#`<C1><86><92><89>][q<F7>x<CE>6w_O"

Your Environment

• Version/GIT-Hash of this module:
  1.0.4

[peteeckel]

Update: It seems that in the meantime all newly added VMs display this problem. It might have something to do with a VMware update the customer installed in the mean time.

What do you need me to test or to provide in terms of debug output in order to fix this problem?

[peteeckel]

Update on this:

I just upgraded to version 1.1.0 of the plugin, the error remains.

When starting the daemon manually I get the error message

Task moRefs failed: Undefined property: stdClass::$overallStatus

Otherwise the system output looks OK. All VMs created after the vSphere was updated to version 6.7 lack a corresponding object in the objects table in the database. All VMs created before have that object (but probably it isn't updated anymore)

I tried to drop and recreate the database, and now all objects are gone and are not recreated by the daemon process. The GUI does not have any Hosts, VMs, Datastores etc. at all anymore. The object table is completely empty.