Change-Id: I04e6c1fca113e3f4cab7cd86045790355c27b61a

This shows up here - https://git.linaro.org/lava/lava-dispatcher.git/about/

Change-Id: I271d56d333efeb60385e11dc814c436057b6ee77

Only top pages are now reachable from the menu (worker, devices, jobs, status
and submit).

Change-Id: If3f55b9b41ab6333d43bdf018c8f499b73b08ad4

Allows to have one url to point users at.

Change-Id: I04d2c1e18a19e932e6581227cd57e4640e19ae93

The old results line are missing some important keys.

Change-Id: Id6f3d748def7d90ef8f8cfc5bfe9acb8e8f4fb92

Change-Id: Ie2f9b465dfae0629101d8f210166ee20931c05d0

Change-Id: I7d0bf0030962e0eff1fd234066096c78e5d082e5

Change-Id: I2729ef361aee395fec981a51834ea77a1cf89705

Bump up the logging level. Add all condition choices in error log.

Change-Id: Id55f295cb2b5ea1507fec6535295776c248e8cf3

The race condition in _save of FileSystemStorage was fixed via
https://code.djangoproject.com/ticket/21602 hence it is redundant to
re-implement it here.

Change-Id: I7cdae9314ce019e18cfaaef16e53b5fef0770587

Crash found by meta-lava.

Change-Id: Iaa910226dda4e36bc81cee521a86226d49d76a77

Resort to using shell comands for decompressing the images and
piping the input data through stdin. If no compression is needed,
do not use pipe but rather only download the file.

Change-Id: Ife5057a51afe1cc124c7f80bf0ed00686761a1c4

Change-Id: I60c43573cc9d861377462298049bed70aaea90a0

Change-Id: Id141e97a67df2975ea12f90ff30a0069ed60fb7f

Change-Id: I2777cc7f476abd4e47d3145166bad6959a1ff38b

Change-Id: Icd5d214bb19cb5c356d69b9e3f4dbf008f9088a7

Change-Id: I3abd6f48541d332f9f508558d90b2fef5d3eaca7

Change-Id: I551a20d33ec8a01c2f4c63d67a5464c0a34b3b36

"POST data should be bytes, an iterable of bytes, or a file object. It cannot
be of type str."

Change-Id: I020fec71769b6ff953428f75692a8681e09645a9

The ci.debian.net autopkgtest environment does not necessarily
enable -kvm, so this test failed. Skip if the existing check
from lava-dispatcher determines that a test job using kvm would
also fail.

Change-Id: Icf13aa15d8bfdd07b73acd01f864b9c6812ae39b

Change-Id: I9e81d32cf4177639e7aa83dd744b7f64cb6a9821

The url *should* start with a "/".

Change-Id: I0262b69d62ed3d47665e76dc13ff93bb56998e2a

This is only available in python3.

Change-Id: I825325e1657ab262131fd049d511bd90bb0f0b54

My BBB was failing to boot a uImage due to an issue with the addresses
(I didn't dive into the details why). Changing the bootm_ addresses to
match the bootz_ addresses (as has been done on various other boards)
solved the issue.

Further more in some cases during auto-login the username was received
as 'suer' rather then 'user', add a small boot character delay to
resolve that.

Change-Id: I84431b7d037623c39c8d25dce1a0d603a3867cd0
Signed-off-by: Sjoerd Simons <sjoerd.simons@collabora.co.uk>

The default u-boot prompt for snow devices is 'snow #', make the prompt
setting more precise in this respect to prevent otherwise lava things
the fdt filename is also a command prompt.

Change-Id: I038348bec90d44b23b658066c9b413328be77e66
Signed-off-by: Sjoerd Simons <sjoerd.simons@collabora.co.uk>

Adjust the bootm address for panda to match the bootz address. This
fixes booting certain kernels/ramdisk combination which would boot fine
with bootz, but didn't work as bootm.

Change-Id: Ie63b8ea62935322be1ba816f0a3b9745e8916b05
Signed-off-by: Sjoerd Simons <sjoerd.simons@collabora.co.uk>

Change-Id: Ie98449b849e5cdd3de6d16a6902750f6f546a146

Our LAB run lots of XNBD jobs, and lots of xnbd-server process were
never killed.

The port allocated is not used for killing, eg
self.parameters['protocols']['lava-xnbd']['port'] is still set to auto.

As suggested by Neil Williams, it's better to "create a dedicated
function which is called from the test job"

So this patch adds a "set_port" function which get the right port
number.

Existing jobs should now add this:
   protocols:
      lava-xnbd:
      - action: nbd-deploy
        request: set_port
after
actions:
- deploy:

Change-Id: I6b464314ac2c1a137162b0a7f955c4501f8a7424
Suggested-by: Jan-Simon Möller <jsmoeller@linuxfoundation.org>

Action mounts the mass storage device for the MPS system and copies
over the contents of the recovery image provided by the test writer.

Change-Id: Ice721a4f84efc5655876d5c41298c0af7d684968

Add an index on TestCase results to improve sql queries speed.
Add links to Recent job errors and Maintenance devices to
job table and device table templates.

Change-Id: Ib6e67b25bf8ae88e20f03ed73230ea7a752f3e71

Change-Id: Ib42bd8165ebc7946bb51fab320f66a365bf0da10

This will raise an infrastructure error if the tool is missing.

Change-Id: Id406dd2576c990c0030a8ca16282ff02535e335a

When a job is really long to cancel we now:
* send a TERM signal (the job starts to cleanup)
* wait 60 seconds
* send a second TERM signal (the job should leave right now)
* wait 60 more seconds
* send a KILL signal

This let more time for the job to send the right lava.job result.

Change-Id: I262b13a3bcc70576fca44127c15d5265864787c3

By default, when stoping a service, systemd will kill all processes started by
this service.
In order to make lava-slave restartable, all lava-run started by lava-slave
should *not* be killed.

Change-Id: I422f71ba70aabbc58cf3241baac1ffedd677d458

Change-Id: Ie45cf3f855190bcbb63c434297b47bf3854c73a2

Change-Id: I8be314e68e9ca47221e9692257c0fbbfae1d73b4

When trying to connect to the master, run some basic DNS checks on
the requested master. Skip sending the HELLO_RETRY unless the
host is at least accessible. Output socket-based error messages
when the master does not resolve. Skip some log calls for IPv6.

Change-Id: I1a3721340f6e4d278dea3e99949fae641d0c1535

This will help reproducing the exact same run.

Also add an xmlrpc api to grab the job configuration.

Change-Id: I64798a12146064d3fb811bb28bf27f49e20fd6d5

Change-Id: I0e077f9d6ea96203d05a0d77ea37814806a10bd0