Hi folks,
The 2022.08 tag has been pushed to master on git.lavasoftware.org.
.deb packages have been built in GitLab CI and are published at
https://apt.lavasoftware.org/release
Docker images for amd64 and arm64 have been built in GitLab CI and
are available from
https://hub.lavasoftware.org/
and
https://hub.docker.com/u/lavasoftware
Changes in this release
==================
## New device-types
New supported devices:
* aaeon-UPN-EHLX4RE-A10-0864
* imx8ulp-evk
* imx93-11x11-evk
* mt8192-asurada-spherion-r0
* synquacer-uboot
## Grub based device-types
Allow device-type to easily disable interrupting boot. This is not required
if grub is build to always drop to a shell.
In the device-type template or device dictionary, add:
```jinja
{% set grub_needs_interrupt = false %}
```
## UUU and BCU
Add support for [BCU](https://github.com/NXPmicro/bcu#readme) to the UUU
based device-types. This utility allows changing the board's boot
configuration (mainly SD card, eMMC or USB Serial Download Protocol)
through a serial interface.
In the device dictionary, add:
```jinja
{% set bcu_board_id = '2-1.3' %}
```
Then in the job definition, you can use bcu directly:
```yaml
- boot:
method: uuu
commands:
- bcu: reset usb
- uuu: -b emmc {boot}
- bcu: set_boot_mode emmc
timeout:
minutes: 20
```
The support has been enabled for imx8dxl-evk imx8ulp-evk imx8mp-evk and
imx93-11x11-evk device-types.
## OpenID connect
LAVA server can now support OpenID connect (for instance Azure ID or
Keycloack) to authenticate users.
In roder to use this feature, you should install `mozilla-django-oidc` and
add to the settings:
```yaml
AUTH_OIDC:
OIDC_RP_CLIENT_ID: "1"
OIDC_RP_CLIENT_SECRET: "bd01adf93cfb"
OIDC_OP_AUTHORIZATION_ENDPOINT: "http://testprovider:8080/openid/authorize
"
OIDC_OP_TOKEN_ENDPOINT: "http://testprovider:8080/openid/token"
OIDC_OP_USER_ENDPOINT: "http://testprovider:8080/openid/userinfo"
```
See [mozilla-django-oidc](
https://mozilla-django-oidc.readthedocs.io/en/stable/settings.html) for the
full list of options.
## Transfer Overlay
For device-type where the base OS does not provide tools to download over
http (like wget or curl), the transfer overlay action can now use NFS.
In order to use this, the job definition would look like:
```yaml
- boot:
[...]
transfer_overlay:
transfer_method: nfs
download_command: mount -t nfs -o nolock
unpack_command: cp -rf
```
Rgds
--
Rémi Duraffort
Tux and LAVA Architect
Linaro
Hi folks,
The 2022.06 tag has been pushed to master on git.lavasoftware.org.
.deb packages have been built in GitLab CI and are published at
https://apt.lavasoftware.org/release
Docker images for amd64 and arm64 have been built in GitLab CI and
are available from
https://hub.lavasoftware.org/
and
https://hub.docker.com/u/lavasoftware
Changes in this release
==================
# Device-types
## New device-types
New supported devices:
* bcm2835-rpi-b-rev2
* jh7100-visionfive
* kontron-bl-imx8mm
* rk3399-roc-pc
# Docker test shell
Starting from this release, when using the docker test shell, the current
device connection will be recorded in the logs as a feedback connection.
This allows to print the DUT logs while running a docker test shell.
# Callbacks
It is now possible to receive a callback notification for both `running`
and `finished` job. You have to set `criteria: all` in the job definition.
# Private instance
It is now possible to require all users to login before accessing any page
outside of the home page, documentation pages and the login page itself by
setting the `REQUIRE_LOGIN` variable in any YAML configuration file under
`/etc/lava-server/settings.d/*.yaml`.
Rgds
--
Rémi Duraffort
TuxArchitect
Linaro
Hi folks,
The 2022.05 tag has been pushed to master on git.lavasoftware.org.
.deb packages have been built in GitLab CI and are published at
https://apt.lavasoftware.org/release
Docker images for amd64 and arm64 have been built in GitLab CI and
are available from
https://hub.lavasoftware.org/
and
https://hub.docker.com/u/lavasoftware
Changes in this release
==================
# Device-types
## New device-types
New supported devices:
* am437x-idk-evm
* am57xx-beagle-x15
* armada-388-clearfog-pro
* hp-14-db0003na-grunt
* imx6dl-udoo
* kontron-kswitch-d10-mmt-6g-2gs
* kontron-kswitch-d10-mmt-8g
* rk3399-khadas-edge-v
* sun9i-a80-cubieboard4
## depthcharge
Add extra_kernel_args parameter to the Depthcharge boot method, to allow
specifying additional kernel arguments specific for each boot action in
multi-stage jobs.
# Debian package
## lava-dispatcher-host
Allow to install the package on Ubuntu focal. This allows to run a lava
worker on Ubuntu focal thanks to `lava-docker-worker`.
# Social accounts
Add a page allowing users to manage the social account connection.
# Performances
## Scheduler
Improve scheduler performance by caching the device-type templates. The
previous caching mechanism was in fact broken.
## Web interface
Improve server performances when browsing the LAVA web interface as
anonymous user.
Rgds
--
Rémi Duraffort
LAVA and TuxArchitect
Linaro
Hi folks,
The 2022.04 tag has been pushed to master on git.lavasoftware.org.
.deb packages have been built in GitLab CI and are published at
https://apt.lavasoftware.org/release
Docker images for amd64 and arm64 have been built in GitLab CI and
are available from
https://hub.lavasoftware.org/
and
https://hub.docker.com/u/lavasoftware
Changes in this release
==================
# Device-types
## New device-types
New supported devices:
* morello
# Authentications
Allow to enable login with multiple social accounts at once. The login page
will display the available options with logos.
# Environment
The dispatcher IP is now available in the test shell actions as
`LAVA_DISPATCHER_IP`.
# Git authentication
LAVA can now clone from authenticated git repositories. The credentials
should be present in the local environment variable, for example via the
`secrets` dictionary. LAVA dispatcher now expands environment variables in
git urls:
```yaml
actions:
- test:
definitions:
- repository: https://${USER}:${TOKEN}@example.com/definitions.git
from: git
[...]
secrets:
USER: MY_USER
TOKEN: MY_TOKEN
```
# lava-docker-worker
The LAVA team advice to replace `lava-worker` by `lava-docker-worker` as
this daemon will automatically upgrade and downgrade the local lava-woker
container to match the remote server version.
For this release, the `lava-docker-worker` daemon stability as been improve
a lot in this release. This is now used in production at Linaro.
# Overlays
The lava dispatcher is now able to run `img2simg` after applying the
overlays. Users should add `spare: true`:
```yaml
rootfs:
url: "http://example.com/rootfs.ext4.xz"
format: ext4
sparse: true
overlays:
modules:
url: "http://example.com/modules.tar.xz"
[...]
```
# Server performances
## logging interval
Starting from this release, the default logging interval is set to 5s.
lava-run will now send the job logs every 5s, instead of every seconds.
This should decrease the load on the LAVA server.
## Scheduler lock
Prior to this release, the lava-scheduler process was locking the worker
table during the scheduling. On large instances, this would prevent the
workers to update their status.
Starting from this release, lava-scheduler is not locking the worker table
anymore.
# XML-RPC API
## Device-types
The XML-RPC API is now returning the `health_denominator` and
`health_frequency` when calling `scheduler.device_types.show`. This
function will also return a boolean (`default_template`) indicating if this
device-type is using a custom templare or the default one.
## Devices
The XML-RPC API allows to change the device-type of a specific device when
calling `scheduler.devices.update`.
## workers
The XML-RPC API call `scheduler.workers.show` is now returning booleans
(`default_config`, `default_env` and `default_env_dut`) to indicate if the
worker is using a custom config (env or env-dut) file.
Rgds
--
Rémi Duraffort
LAVA and Tux Architect
Linaro
Hi folks,
The 2022.03 tag has been pushed to master on git.lavasoftware.org.
.deb packages have been built in GitLab CI and are published at
https://apt.lavasoftware.org/release
Docker images for amd64 and arm64 have been built in GitLab CI and
are available from
https://hub.lavasoftware.org/
and
https://hub.docker.com/u/lavasoftware
Changes in this release
==================
# Device-types
## New device-types
New supported devices:
* at91sam9g20ek
# Server performances
Improve server performances by dropping COUNT queries that where used for
pagination. Also cache permissions to avoid recalculation in the same page.
# lava-(docker)-worker
Allow to set `--job-log-interval` for both `lava-docker-worker` and
`lava-worker`. On large instances, admins can increase the value to lower
server load.
# lava-dispatcher-host
Fix a file descriptor leak. Without this patch, after some jobs, the
process will run out of file descriptors.
# lava-run
Add job definition secrets to the environment when starting `lava-run`.
This allows to download from private git repositories by providing
authentication in the job definition secrets.
Rgds
--
Rémi Duraffort
TuxArchitect
Linaro
Hi folks,
The 2022.02 tag has been pushed to master on git.lavasoftware.org.
.deb packages have been built in GitLab CI and are published at
https://apt.lavasoftware.org/release
Docker images for amd64 and arm64 have been built in GitLab CI and
are available from
https://hub.lavasoftware.org/
and
https://hub.docker.com/u/lavasoftware
Changes in this release
==================
# Device-types
## New device-types
New supported devices:
* seco-c61
## d05
Use NFS version 3 by default when loading the root file-system with NFS.
## dephtcharge
Allow to boot from emmc. Typical extra kernel arguments for booting a
Chrome OS image are `root=/dev/mmcblk0p3 cros_debug cros_secure`.
## hp-x360-14-G1-sona
Automatically retry on boot failures to workaround bootloader issues.
## UUU
Enable "on worker" uuu_usb_otg_path configuration instead of modifying
devices definitions located on the master.
Before this change, the LAVA device dictionary contained static uuu USB ID
as a string, like:
```jinja
{% set uuu_usb_otg_path = '1:324' %}
```
The dictionary can now contain a command that will actually return this
string.
```jinja
{% set uuu_usb_otg_path_command = ['board-control', 'imx8mq-evk-01',
'extra', 'uuu_otg_path', '--silent'] %}
```
# docker + adb,fastboot
We often observe race conditions where a USB device will disconnect between
the time when it gets added to the docker command line as a `--device=
option` and the time that the container actually starts.
Instead of running one-off containers, start the containers first, wait for
them to be up, map devices into them with lava-dispatcher-host, and then
run the command. This way, if the device gets disconnected it will be
re-shared with the container once it is enumerated again.
# Settings
The LAVA server settings can now be configured from environment variables.
Every variables that starts with `LAVA_SETTINGS_` will be added to the
django settings.
For complex settings, admins can use LAVA_JSON_SETTINGS which is expected
to be a dictionary as json, base64 encoded. For instance:
```python
import base64
import json
data = {
"WORKER_AUTO_REGISTER_NETMASK": ["::1"]
}
print(base64.b64encode(json.dumps(data).encode("utf-8")).decode("utf-8"))
```
In order to use it, add to the environment:
`LAVA_JSON_SETTINGS="eyJXT1JLRVJfQVVUT19SRUdJU1RFUl9ORVRNQVNLIjogWyI6OjEiXX0="`.
See [the documentation](
https://lava.readthedocs.io/en/latest/admin/basic-tutorials/instance/config…
).
# Container images
Fix LDAP support when using LDAPS and upgrade sentry-sdk to 1.5.5 in
lava-server images.
# Debian packaging
Ensure `lava-celery-worker`, `lava-dispatcher-host` and
`lava-docker-worker` are restarted on upgrades.
# Worker
## Logging interval
By default, lava-run will send the new logs to the server every second. For
large labs, this would generate a consequent load on the server. This
interval is now configurable when calling `lava-docker-worker` and
`lava-worker`.
Rgds
--
Rémi Duraffort
TuxArchitect
Linaro
Hi folks,
The 2022.01.2 tag has been pushed to master on git.lavasoftware.org.
.deb packages have been built in GitLab CI and are published at
https://apt.lavasoftware.org/release
Docker images for amd64 and arm64 have been built in GitLab CI and
are available from
https://hub.lavasoftware.org/
and
https://hub.docker.com/u/lavasoftware
This release is reverting the "Balance job scheduling between submitters"
feature. In fact on large instances with large queue, this feature is
making scheduling way slower.
The feature will be reworked and reintroduced in a later release when the
performance issue has been fixed.
Rgds
--
Rémi Duraffort
TuxArchitect
Linaro
Hi folks,
The 2022.01.1 tag has been pushed to master on git.lavasoftware.org.
.deb packages have been built in GitLab CI and are published at
https://apt.lavasoftware.org/release
Docker images for amd64 and arm64 have been built in GitLab CI and
are available from
https://hub.lavasoftware.org/
and
https://hub.docker.com/u/lavasoftware
Changes in this release
==================
# Device-types
## New device-types
New supported devices:
* acer-cb317-1h-c3z6-dedede
* dell-latitude-5400-4305U-sarien (split from dell-latitude-5400-sarien)
* fsl-ls1046a-frwy
* fsl-ls2088a-rdb
* fsl-lx2162a-qds
* hp-x360-12b-ca0500na-n4000-octopus
* lenovo-TPad-C13-Yoga-zork
* rk3399-rock-pi-4b
* seco-b68
*
## fsl-l*-rdb
Allow to boot from an alternate bank. In order to use this feature, set
`uboot_altbank` to `True` in the job context:
```yaml
context:
uboot_altbank: true
```
Admins should set `uboot_altbank_cmd` in the device-type template.
# lava-dispatcher-host
Add support for docker device sharing under cgroups v2.
Under cgroups v2, device access control is done with BPF programs only.
When docker creates a container, it already attaches a BPF program to that
container cgroup. lava-dispather-host replaces that BPF program with one of
its own, that allows the regular list of devices containers can usually
access (`/dev/null`, `/dev/zero`, ...), plus all the devices shared with
the container. Subsequent device sharing with the same container overrides
that BPF program with a new one
# Scheduler
## Balance job scheduling between submitters
This patch is intended to prevent submitter starvation when jobs have the
same priority and single submitter sends a large batch of jobs.
It is implemented by determining a number of already running jobs for each
submitter and using it to annotate the jobs submitted to the queue. This
allows reordering the job queue for picking a job from the submitter that
could potentially be starving for an available device. To ensure fairness
in edge cases (equal number of running jobs, single device of given device
type) jobs sharing the same set of attributes are randomised.
## Active workers
Only schedule jobs on active workers. When a worker is in maintenance, no
jobs will be scheduled on the attached devices.
# kexec
Add an option to download the artifacts from deployment action. The job
would look like:
```yaml
- deploy:
to: download
images:
kernel:
url: https://example.com/Image.bin
initrd:
url: https://example.com/rootfs.cpio.gz
compression: gz
- boot:
method: minimal
auto_login:
[...]
prompts:
- "root@localhost:~# "
- boot:
method: kexec
deploy: true
kernel: /boot/Image.bin
initrd: /boot/rootfs.cpio.gz
command: /usr/sbin/kexec
[...]
```
# Overlay
Add support for tar filesystem. This allow to add overlays to nfsrootfs for
the boards that use NFS.
Rgds
--
Rémi Duraffort
LAV A and Tux Architect
Linaro
Hi folks,
The 2021.11 tag has been pushed to master on git.lavasoftware.org.
.deb packages have been built in GitLab CI and are published at
https://apt.lavasoftware.org/release
Docker images for amd64 and arm64 have been built in GitLab CI and
are available from
https://hub.lavasoftware.org/
and
https://hub.docker.com/u/lavasoftware
Changes in this release
==================
# Device-types
## New device-types
New supported devices:
* asus-cx9400-volteer
* lenovo-hr330a-7x33cto1ww-emag
* sc7180-trogdor-lazor-limozeen
## kontron-sl28
Define `initrd_high` and `fdt_high` as these boards are 64 bits.
## grub devices
### bootp interface
It's now possible to set the `bootp` interface when net booting with grub.
Admins should add to device-dict:
```jinja
{% set net_interface = "efinet2" %}
```
### base_kernel_args
Allow to set `extra_kernel_args` in grub jobs. This was impossible in
previous version due to a bug in the jinja templates.
This can be set in the device dictionary or the job definition context.
### FVP
The support for reading feedback channels when running an FVP model as been
improved thanks to the work of Paul Sokolovsky.
With these changes LAVA is able to consume and record most of the feedback
channels data.
Job submitter can also configure the timeout on feedback channels reading
in the finalize action:
```yaml
timeouts:
connections:
read-feedback: 20
```
With a large timeout, LAVA will have more time to consume the remaining
data when finalizing a job.
## Minimal boot
Add support for `pre_os_command` and `pre_power_command` in minimal boot
action.
The job definition would look like:
```yaml
- boot
method: minimal
pre_power_command: true
```
These commands can be useful to activate or deactivate external hardware
before powering on the DUT.
## NBD
In order to boot DUT with NBD support, LAVA was using `xnbd-server`.
This project is not maintained anymore so LAVA is now using `nbd-server`
instead.
# Management command
Admins can now list jobs with `lava-server manage jobs list`.
To list health check jobs than where submitted in the last 10 days:
```shell
lava-server manage jobs list --submitter lava-health --newer-than 10d
```
To run lxc jobs, that where submitted in the la 10 days:
```shell
lava-server manage jobs list --newer-than 10d --lxc
```
# Performances
The performances of the following pages has been improved drastically:
* /results
* /scheduler/mydevices
* /scheduler/mydevicetypehealthhistory
Rendering these pages will generate way fewer SQL queries.
Rgds
--
Rémi Duraffort
LAVA and TuxArchitect
Linaro
Hi folks,
The 2021.10 tag has been pushed to master on git.lavasoftware.org.
.deb packages have been built in GitLab CI and are published at
https://apt.lavasoftware.org/release
Docker images for amd64 and arm64 have been built in GitLab CI and
are available from
https://hub.lavasoftware.org/
and
https://hub.docker.com/u/lavasoftware
Changes in this release
==================
# Device-types
## New device-types
New supported devices:
* arduino-nano-33-ble
* imx8mp-ab2
* meson-sm1-odroid-c4
* pc-k10n78
## mt8192-asurada
mt8192-asurada as been renamed mt8192-asurada-rev1 to match the device-tree
name.
## FVP
In order to fix some issues with the secondary consoles that should be read
at all time, LAVA is now using a specific wrapper called `lava-outerr`.
The wrapper script consumes the sub-process std(out|err) as fast as
possible, keep the data in a buffer and written to std(out|err) whenever
possible.
## nrf52-nitrogen
Allow to set connect_before_flash. The value is False for backward
compatibility. You can set it in the device dictionary:
```jinja
{% set connect_before_flash = true %}
```
# External authentication
Since [2021.03](2021.03#authentication), LAVA can delegate the
authentication to GitLab using the `django-allauth` module.
This release add the possibility to use any supported [provider](
https://django-allauth.readthedocs.io/en/latest/providers.html).
In order to use it, add this to the configuration:
```yaml
AUTH_SOCIALACCOUNT: "{'gitlab':{'GITLAB_URL': 'https://gitlab.example.com
'}}"
```
See `django-allauth` documentation for the right parameters.
For background compatibility, `AUTH_GITLAB_URL` and `AUTH_GITLAB_SCOPE` are
still supported.
# Interactive tests
When running tests with LAVA test shell, LAVA will regularly read the
connections from the non-active namespaces. In the logs, this is called
"feedbacks". This is important for some device-types like fvp models.
Interactive tests are now also regularly reading the non-actives
connections to fix some issues with the fvp device-type.
# kexec
The kexec boot method is now able to crash the kernel to test the kdump
feature.
The job definition will look like:
```yaml
boot:
method: kexec
kernel: /home/vmlinux
dtb: /home/dtb
initrd: /home/initrd
command: /sbin/kexec
on_panic: true
boot_message: "Starting crashdump kernel"
prompts:
- 'root@debian:~#'
```
LAVA will use `--load-panic` instead of `--load` and will call `echo c >
/proc/sysrq-trigger` to force a kernel crash.
# Overlays
It's now possible to add single files to a given root file system (or
initrd) with the overlays feature.
The job definition should look like:
```yaml
rootfs:
image_arg: "-initrd {rootfs}"
url: "https://example.com/rootfs.cpio.zst"
compression: zstd
format: cpio.newc
overlays:
onefile:
url: https://example.com/file
format: file
path: /hello-world
```
# qemu and docker
The qemu-system packages included in the lava-dispatcher docker image are
now coming from bullseye-backports.
Rgds
--
Rémi Duraffort
LAVA and Tux Architect
Linaro
