Custom field in custom script does not save to database

[MalfuncEddie]

Hi,

I am kind of stuck on why the following piece of custom script code does not update the VirtualMachine.
When running I, I get the following log loglines

Script Log

Line	Level	Message
1	Info	updating testvm
2	Info	field : None

            vms = VirtualMachine.objects.filter(name=vmName)
            if vms.count() == 1:
                if (not vms[0].custom_field_data["updated"]):
                    self.log_info("updating "+ vmName)
                    vms[0].custom_field_data["updated"] = True
                    vms[0].full_clean()
                    vms[0].save()
                    self.log_info("field :" + str(vms[0].custom_field_data["updated"]))

                else:
                    self.log_info("update is already True")

            elif vms.count() == 0:
                self.log_info("0 vm's found ignoring " + vmName)
            else:
                self.log_info("More than 1 vm found ignoring " +vmName)

I do get a change log update but nothing is changed. the custom field updated is still None.
So I detect it needs to update the field but vms[0].save() does nothing?

[candlerb]

Just to check the obvious: you have created a Custom Field definition called "updated" in the Netbox UI, and set its data type to "boolean"? And you've associated this Custom Field definition with content type "Virtualization > virtual machine" ?

(Although if you hadn't done all that, the "clean" operation should have raised an exception)

And what Netbox version are you running?

[candlerb]

First I had to fill in the missing bits to make a runnable script:

from virtualization.models import VirtualMachine
from extras.scripts import Script, StringVar

class TestVM(Script):
    class Meta:
        name = "Test VM"
        description = "Test VM custom field"

    vm_name = StringVar(label="VM name")

    def run(self, data, commit):
            vmName = data["vm_name"]
            vms = VirtualMachine.objects.filter(name=vmName)
            if vms.count() == 1:
                if (not vms[0].custom_field_data.get("updated")):
                    self.log_info(f'updating {vmName}')
                    self.log_info(f'field : {vms[0].custom_field_data.get("updated")}')
                    vms[0].custom_field_data["updated"] = True
                    self.log_info(f'field : {vms[0].custom_field_data.get("updated")}')    # NOTE THIS!
                    vms[0].full_clean()
                    vms[0].save()
                    self.log_info(f'field : {vms[0].custom_field_data.get("updated")}')

                else:
                    self.log_info("updated is already True")

            elif vms.count() == 0:
                self.log_info("0 vm's found ignoring " + vmName)
            else:
                self.log_info("More than 1 vm found ignoring " +vmName)

What's clear from this is that Django fetches a fresh dict every time. See the output of "NOTE THIS!"

The problem appears to be: vms is a proxy object (a query set), and every time you write vms[0] you are asking it to fetch a fresh copy of the object from the database.

To get around this, you can crystallize the query into real local objects, which you can do with list(...)

    def run(self, data, commit):
            vmName = data["vm_name"]
            vms = list(VirtualMachine.objects.filter(name=vmName))
            if len(vms) == 1:
                if (not vms[0].custom_field_data.get("updated")):
                    self.log_info(f'updating {vmName}')
                    self.log_info(f'field : {vms[0].custom_field_data.get("updated")}')
                    vms[0].custom_field_data["updated"] = True
                    self.log_info(f'field : {vms[0].custom_field_data.get("updated")}')
                    vms[0].full_clean()
                    vms[0].save()
                    self.log_info(f'field : {vms[0].custom_field_data.get("updated")}')

                else:
                    self.log_info("updated is already True")

            elif len(vms) == 0:
                self.log_info("0 vm's found ignoring " + vmName)
            else:
                self.log_info("More than 1 vm found ignoring " +vmName)

Or maybe just vm = vms[0] would be sufficient.

However, I think it would be better to simplify this script using get():

vm = VirtualMachine.objects.get(name=vmName)

(which will give an exception if zero or multiple objects match). Even better, use Netbox's ObjectVar, which will allow the user to select a VM interactively from a drop-down list.

[MalfuncEddie]

Thank you so much!

[MalfuncEddie]

Even better, use Netbox's ObjectVar, which will allow the user to select a VM interactively from a drop-down list.

Well actually it is a textfield where there is a multiline list of vm's that need to be updated. But since the names from 3rd party sources not necessarily match I wrote a method that first tries to find it by uuid, then by name or tenant ... (yes a lot of regex/filters
... why do you ask ;))

[MalfuncEddie]

Yes I added it, I also did a test with a vm that has (via GUI) the value True . Then I go to the self.log_info("update is already True") part
So it is a "saving" issue. Am I correct by your remark that this should work?

version v3.2.2

[MalfuncEddie]

I do get a

Warning: Comparing non-atomic change to previous change record

In the changelog ( Difference: No Changes)