1. Make sure VMs are started after dom0 actual memory usage is reported
to qmemman, otherwise dom0 will hold 4GB, even if just a little over 1GB
is needed at that time.
2. Request only vm.memory MB from qmemman, instead of vm.maxmem. While
HVM with PCI devices indeed do not support populate-on-demand, this is
already handled in libvirt XML.
The later may often cause VM startup fail on systems with 8GB of memory,
because maxmem is 4GB there and with dom0 keeping the other 4GB (see
point 1) there is not enough memory to start any sych VM.
FixesQubesOS/qubes-issues#3462
The service needs to be started after qubesd and qubes-qmemman to be
able to start domains. The same applies on shutdown - should be ordered
before shutting down qubesd.
qubesd service is critical for Qubes usage, so even in case of critical
error crashing the whole service, make sure it is restarted.
Set delay to 1s (default 100ms), to allow other services to restart too,
if crash was caused by some other service (like libvirtd crash).
QubesOS/qubes-issues#2960
Since qubesd properly handle chained startup of sys-net->sys-firewall
etc, we don't need a separate service to start netvm explicitly earlier.
FixesQubesOS/qubes-issues#2533
This service currently does more harm (desync libvirt state with
reality) than good. Since we have qubesd, we may come back to
implementing it properly using libvirt events.
Prior to this commit, the qubes-core.service inherited systemd's default
timeout value of 90 seconds. With slow hard disk drives, this caused the
dom0 shut-down sequence to proceed even if some VMs were still not fully
shut down at the time of dom0 shut down.
This commit aims to avoid this issue by setting the service stop timeout to
180 seconds.
Signed-off-by: M. Vefa Bicakci <m.v.b@runbox.com>
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
systemd-user-sessions.service is specicically for that, do not use hack
(plymouth-quit.service), which doesn't work when the service is
disabled.
FixesQubesOS/qubes-issues#1250
It may happen (especially when VM doesn't close cleanly and needs to be
killed) that qubesdb-daemon will not notice VM shutdown immediately.
Normally it would stop after 60s timeout, but speed it up in case of
system shutdown
QubesOS/qubes-issues#1425
This is part of fixing qvm-start.
qmemman was moved with minimal touching, mainly module names.
Moved function parsing human-readable sizes from core2. This function is
wrong, because it treats k/M/G as 1024-based, but leave it for now.
When user logins, login script will try to connect all guid to all the
running VMs. If VMs are still booting at this stage, will never
automatically get its guid (until user tries to start some program
there). This can for example lead to lack of nm-applet icon.
Mostly done. Things still using xenstore/not working at all:
- DispVM
- qubesutils.py (especially qvm-block and qvm-usb code)
- external IP change notification for ProxyVM (should be done via RPC
service)
Forking daemon after initializing hypervisor connection can cause
problems (and actually does in case of libvirt).
To notify systemd when daemon is ready use notify socket (previously it
was termination of parent process).
It is common for both dom0 and VM, and also quite linux-specific
(other OSes will need other implementation). So move to linux-specific
repo (not dom0-specific).
