Pickling pyvmomi class using multiprocessing

[welibekov]

Hi folk, I try do use python multiprocessing module to parallelize some data gathering from vcenter. Unfortunately i have faced with following problem during starting parallel processes:

_pickle.PicklingError: Can't pickle <class 'pyVmomi.VmomiSupport.vim.VirtualMachine'>: attribute lookup vim.VirtualMachine on pyVmomi.VmomiSupport failed

So the question is, it's that true, that pymomi doesn't support "right pickling"? Thanks, and appreciate.

[george-gwu]

I ran into this same issue but found out that there is a JSON encoder that can be used to move these objects between processes.

from pyVmomi.VmomiSupport import VmomiJSONEncoder

Use this to create your list: jsonSerialized= json.dumps(pfVmomiObj, cls=VmomiJSONEncoder)

Then in the mapped function, use this to recover the object: pfVmomiObj = json.loads(jsonSerialized)

[cube0x8]

VmomiJSONEncoder solved the problem and I can pass a serialized vim.VirtualMachine object to the mapped function.

But how do I do the opposite conversion (from json to vim.VirtualMachine object)? This: pfVmomiObj = json.loads(jsonSerialized) returns a dict, so I cannot use pfVmomiObj with APIs that take a ManagedEntity as argument (for example InitiateFileTransferToGuest).

Thanks