Process exit is too late for long-running processes. Especially tests,
which create a lot of such objects which leads to depleting all
the file descriptors.
QubesOS/qubes-issues#1697
@woju what was consensus on 'servicevms' dir? Since "being network
provider" is no longer separate type and can be changed during VM
lifetime, maybe we should abandon that idea? Or maybe just set it as VM
property (some storage pool config argument) - that way mgmt stack could
put all the sys-* VMs into separate directory, regardless of the type.
The following list is bollocks. There were many, many more.
Conflicts:
core-modules/003QubesTemplateVm.py
core-modules/005QubesNetVm.py
core/qubes.py
core/storage/__init__.py
core/storage/xen.py
doc/qvm-tools/qvm-pci.rst
doc/qvm-tools/qvm-prefs.rst
qubes/tools/qmemmand.py
qvm-tools/qvm-create
qvm-tools/qvm-prefs
qvm-tools/qvm-start
tests/__init__.py
vm-config/xen-vm-template-hvm.xml
This commit took 2 days (26-27.01.2016) and put our friendship to test.
--Wojtek and Marek
Force refreshing domain list after starting new VM, even if the
triggering watch wasn't about domain list change. Otherwise (with
outdated domain list) memory allocated to a new VM, but not yet used may
be redistributed (leaving to little to the Xen itself).
FixesQubesOS/qubes-issues#1389
Retrieve a domain list only after obtaining global lock. Otherwise an
outdated list may be used, when a domain was introduced in the meantime
(starting a new domain is done with global lock held), leading to #1389.
QubesOS/qubes-issues#1389
Test if DispVM firewall rules are:
- inherited from a calling VM
- saved properly (so will be effective after firewall reload)
- saved only for that DispVM (#1608)
QubesOS/qubes-issues#1608
It should be created at VM creation time (or template changes commit).
But for example for HVM templates created before implementing
QubesOS/qubes-issues#1573, there would be no such image. So create it
when needed, just before VM startup
FixesQubesOS/qubes-issues#1602
In this case, qmemman would not manage to get any memory state for such
VM, so memory_current would be 'None'. But later it would perform
arithmetics on it, which would result in qmemman crash (TypeError
exception).
FixesQubesOS/qubes-issues#1601
Regression test for:
QubesOS/qubes-issues#1515
It assumes existing broken symlink in user dom0 home. Thanks for systemd
it is the case (`~/.local/share/systemd/user`).
We still can't support running HVM template and its VMs simultaneously
(easily), but still, have root-cow.img handled for HVM template, to
allow qvm-revert-template-changes.
FixesQubesOS/qubes-issues#1573
When VM got some memory assigned, balloon driver may not pick it up
immediatelly and the memory will still be seen as "free" by Xen, but VM
can use (request) it at any time. Qmemman needs to take care of such
memory (exclude it from "free" pool), otherwise it would redistribute it
to other domains, allowing the original domain to drain Xen memory pool.
Do this by redefining DomainState.memory_actual - it is now amount of
memory available to the VM (currently used, or possibly used). Then
calculate free memory by subtracting memory allocated but not used
(memory_target-memory_current).
FixesQubesOS/qubes-issues#1389
This is needed, because when tests fail to load, unittest module
generates artificial „LoadTestsFailure” class -- not based on our
QubesTestCase, so it does not have .log attribute.
