Refreshing Qube Manager Immediately after closing DispVM causes varied critical errors

[esote]

Qubes OS version:

R4

Affected component(s):

Qube Manager

---

Steps to reproduce the behavior:

1. Launch a disposable virtual machine. In my case, I open Firefox from fedora-26-dvm.
2. Close Firefox (thus shutting down the virtual machine), while simultaneously refreshing Qube Manager.

Expected behavior:

No crash and critical error upon refreshing Qube Manager directly after closing a DispVM.

Actual behavior:

Refreshing Qube Manager right after closing a DispVM (such as closing the last window open, thus shutting it down), causes different critical errors to pop-up:

Sometimes I get this one

Whoops. A critical error has occured. This is likely a bug in Qubes Manager.
QubesPropertyAccessError: Failed to access 'label' property
at line 888
of file /usr/lib/python3.5/site-packages/qubesmanager/qube_manager.py.

----
line: raise qubesadmin.exc.QubesPropertyAccessError(item)
func: __getattr__
line no.: 203
file: /usr/lib/python3.5/site-packages/qubesadmin/base.py
----
line: self.value = 5 + vm.label.index
func: get_vm_icon
line no.: 106
file: /usr/lib/python3.5/site-packages/qubesmanager/table_widgets.py
----
line: (icon_path, tooltip) = self.get_vm_icon(vm)
func: __init__
line no.: 86
file: /usr/lib/python3.5/site-packages/qubesmanager/table_widgets.py
----
line: self.type_widget = table_widgets.VmTypeWidget(vm)
func: __init__
line no.: 81
file: /usr/lib/python3.5/site-packages/qubesmanager/qube_manager.py
----
line: vm_row = VmRowInTable(vm, row_no, self.table)
func: fill_table
line no.: 427
file: /usr/lib/python3.5/site-packages/qubesmanager/qube_manager.py
----
line: self.fill_table()
func: update_table
line no.: 465
file: /usr/lib/python3.5/site-packages/qubesmanager/qube_manager.py
----
line: self.update_table()
func: action_refresh_list_triggered
line no.: 888
file: /usr/lib/python3.5/site-packages/qubesmanager/qube_manager.py

This critical error has happened twice now, both when refreshing immediately after closing a DispVM.

I have also gotten this one, when doing the same action (perhaps the timing is different, but nonetheless).

Whoops. A critical error has occured. This is most likely a bug in Qubes Manager.
QubesPropertyAccessError: Failed to access 'qid' property
at line 77
of file /usr/lib/python3.5/site-packages/qubesmanager/table_widgets.py.

With the details below:

----
line: raise qubesadmin.exc.QubesPropertyAccessError(item)
func: __getattr__
line no.: 203
file: /usr/lib/python3.5/site-packages/qubesadmin/base.py
----
line: if self.vm.qid == 0:
func: __lt__
line no.: 77
file: /usr/lib/python3.5/site-packages/qubesmanager/table_widgets.py

Immediately following the error shown above (the one with line 77, not 888), I get about a dozen of these errors (each identical, I suppose it has to do with the number of VMs I have).

Whoops. A critical error has occured. This is most likely a bug in Qubes Manager.
QubesPropertyAccessError: Failed to access 'qid' property
at line 79
of file /usr/lib/python3.5/site-packages/qubesmanager/table_widgets.py.

----
line: raise qubesadmin.exc.QubesPropertyAccessError(item)
func: __getattr__
line no.: 203
file: /usr/lib/python3.5/site-packages/qubesadmin/base.py
----
line: elif other.vm.qid == 0:
func: __lt__
line no.: 79
file: /usr/lib/python3.5/site-packages/qubesmanager/table_widgets.py

General notes:

In the error "occurred" is misspelled as "occured" (should have two r's). There is also inconsistent naming between "Qubes Manager" and "Qube Manager."

---

Related issues:

None that i have found.

[underthings]

Just would like confirm same issue.

[underthings]

Just would like confirm same issue.

[mkow]

Similar issue here, with a different error.

Shutting down a template VM (using Qubes Manager) and then clicking "Refresh" a few times resulted in the following error:

Whoops. A critical error has occured. This is most likely a bug in Qubes Manager.
RuntimeError: wrapped C/C++ object of type VmNameItem has been deleted at line 888 of file /usr/lib/python3.5/site-packages/qubesmanager/qube_manager.py.

Details:

----
line: self.vms_in_table[selected_qid].name_widget)
func: fill_table
line no.: 437
file: /usr/lib/python3.5/site-packages/qubesmanager/qube_manager.py
----
line: self.fill_table()
func: update_table
line no.: 465
file: /usr/lib/python3.5/site-packages/qubesmanager/qube_manager.py
----
line: self.update_table()
func: action_refresh_list_triggered
line no.: 888
file: /usr/lib/python3.5/site-packages/qubesmanager/qube_manager.py

[mkow]

Similar issue here, with a different error.

Shutting down a template VM (using Qubes Manager) and then clicking "Refresh" a few times resulted in the following error:

Whoops. A critical error has occured. This is most likely a bug in Qubes Manager.
RuntimeError: wrapped C/C++ object of type VmNameItem has been deleted at line 888 of file /usr/lib/python3.5/site-packages/qubesmanager/qube_manager.py.

Details:

----
line: self.vms_in_table[selected_qid].name_widget)
func: fill_table
line no.: 437
file: /usr/lib/python3.5/site-packages/qubesmanager/qube_manager.py
----
line: self.fill_table()
func: update_table
line no.: 465
file: /usr/lib/python3.5/site-packages/qubesmanager/qube_manager.py
----
line: self.update_table()
func: action_refresh_list_triggered
line no.: 888
file: /usr/lib/python3.5/site-packages/qubesmanager/qube_manager.py

[donob4n]

Hi,

I think this bug is due 'app.domains' cache, like a similar problem which I had: #3875

When you close the dispVM is removed so domains needs to call a clear_cache() or it will try to get the removed VM.

In current version it should work just calling clear_cache() when doing a manual Refresh.

[donob4n]

Hi,

I think this bug is due 'app.domains' cache, like a similar problem which I had: #3875

When you close the dispVM is removed so domains needs to call a clear_cache() or it will try to get the removed VM.

In current version it should work just calling clear_cache() when doing a manual Refresh.

[marmarta]

Should be fixed by QubesOS/qubes-manager#91

[marmarta]

Should be fixed by QubesOS/qubes-manager#91