Beginning of a proper test suite, using qemu. Test just runs docker
for now, will add further integration tests.
Signed-off-by: Justin Cormack <justin.cormack@docker.com>
This is less to do with installing modules (which we generally don't expect to
use in Moby) but to populate /lib/modules/`uname -r`/modules.builtin which
turns:
moby:~# modprobe ip_vs
modprobe: FATAL: Module ip_vs not found in directory /lib/modules/4.4.14-moby
moby:~# modprobe nf_nat
modprobe: FATAL: Module nf_nat not found in directory /lib/modules/4.4.14-moby
moby:~#
into:
moby:~# modprobe ip_vs
moby:~# modprobe nf_nat
moby:~#
which reduces the amount noise in the logs, e.g. in docker.log:
time="2016-07-04T11:21:58Z" level=warning msg="Running modprobe nf_nat failed with message: `modprobe: WARNING: Module nf_nat not found in directory /lib/modules/4.4.14-moby`, error: exit status 1"
A fair number of these appear in the logs.
This also stops various tools logging about /lib/modules/`uname -r` not
existing (there was one in the boot log until recently I think)
Signed-off-by: Ian Campbell <ian.campbell@docker.com>
This is mac only (for now) and will not actually do anything until syslogd is
told to forward to /var/run/syslog.vsock.
syslog uses a SOCK_DGRAM connection to /var/run/syslog.vsock, however vsock
today is SOCK_STREAM only, so we need to "packetise" the stream. Do so by
writing the datagram length as a (little-endian) uint32 before the data itself.
This is slightly modelled after rfc6587 (syslog over TCP) but simplified by
using a 4-byte binary value rather than ASCII digits.
Arrange for vsudd to start before the logger so it is ready and waiting.
Note that the code in vsyslog.go needs to be rather careful about its own
logging, in particular logging forwarding failures over syslog seems likely to
make things worse. Instead this file logs to the console when errors occur,
this will be captured by the logging of the hyperkit VM console.
Signed-off-by: Ian Campbell <ian.campbell@docker.com>
Reconfigure chronyd in a udhcpc post-bound hook and restart it, but use
conditionalrestart so it is not started if it isn't running, which is the case
during boot when this hook happens from init.d/networking via ifup, running
"service ... start" in that context results in the daemon running but the
service db saying it isn't.
Doing it in the udhcpc hook leaves a nice hole where support for getting the
NTP server via DHCP can fit in once it is available.
This also switches from using "initstepslew" to "makestep", the former was
responsible for the long delay starting chronyd (meaning we can now drop
"before chronyd" from Docker's init dependencies), while the later will force a
step if the time is out by more than the configured amount (half a second
here), meaning it will be corrected on resume. The rest of the time it chrony
will do the usual slewing to keep the clock accurate. Passing -1 as the limit
to "makestep" ensures the step will always occur, if it were positive it would
only be done that number of times.
We add "iburst" and "minpoll 2" to the default configuration file to speed up
initial sync and resync after sleep respectively, resync after sleep happens in
a couple of seconds. In "ntp=gateway" mode we also patch in "trust", this isn't
in the default config file since we won't be so sure about DHCP provided
servers, but when talking to a host provided NTP we want to use its time (for
sync accross host mounts) even if it has a crazy idea about what the time is.
With this we can also remove the 15min ntp sync, which in turn gets rid of the
need for cron.
Signed-off-by: Ian Campbell <ian.campbell@docker.com>
On a Hyper-V system we can only register one listening endpoint (with
a GUID), so we need to accept connections, read a header and then
start the proxy.
If the binary has argv[0] == "proxy-vsockd" then run this new frontend.
Signed-off-by: David Scott <dave.scott@docker.com>
The tap-vsockd service is started from eth1's pre-up script.
On the Mac: the bind of the AF_HYPERV socket fails, and eth1 is skipped.
On Windows: if there is no database then eth1 is skipped.
On Windows: if the database has network=native then eth1 is brought up.
Signed-off-by: David Scott <dave.scott@docker.com>
This will make it easier to add a vsock interface as well as a regular
TCP/IP one.
As a side-effect, the server is now able to handle more than one concurrent
request!
Signed-off-by: David Scott <dave.scott@docker.com>
This is a bit ugly as we copied and then modified the chronyd init.d
script as shipped in the package.
Signed-off-by: Rolf Neugebauer <rolf.neugebauer@docker.com>
- Don't start hv_fcopy_daemon. It seems to exit and we are not
using it anyway. Also, remove it from the initrd
- Move hv_kvp_daemon and hv_vss_daemon to start earlier
(before networking and docker being started)
Signed-off-by: Rolf Neugebauer <rolf.neugebauer@docker.com>
LLMNR is the Windows equivalent of mDNS.
Import the LLMNR daemon from https://github.com/tklauser/llmnrd
It needed some small modification to compile cleanly on Moby
LLMNRD is only started when running inside a Hyper-V VM
Signed-off-by: Rolf Neugebauer <rolf.neugebauer@docker.com>
These are WIP taken from git@github.com:stefanha/linux.git#vsock
(==4c9d2a6be1c6, using "cherry-pick -x") and correspond to RFC v5 of the
frontend patches posted in
http://thread.gmane.org/gmane.linux.kernel.virtualization/27455
There is no corresponding spec proposal update yet, but this set of patches
correspond (roughly) to addressing the feedback on v4 of the spec proposal
http://thread.gmane.org/gmane.comp.emulators.virtio.devel/1062.
kernel_config.arm modifications copied from x86, not tested.
Added /etc/kernel-patches/ directory to the image to be consumed by the
licensing.
Signed-off-by: Ian Campbell <ian.campbell@docker.com>
