We have a requirement for some test cases that the artifacts need to be modified (negative scenario) in order
to verify some scenarios. we are currently using postprocess under deploy action for this purpose.
As of part postprocess, we need to mount partitions of disk image and make some changes based on the requirement.
However, we are unable to mount the partitions using losetup command.
docker:
image: debian:bookworm
steps:
- apt-get update
- loop_dev=$(losetup -P -f --show image.wic)
- echo $loop_dev
The job failed with below error,
losetup -P -f --show image.wic
losetup: cannot find an unused loop device
It looks like the docker container needs root privileges. It works when we run locally with "privileged:true".
Is there any way we can pass "privileged: true" option to docker image through LAVA job definitions?
Any inputs regarding this matter are highly appreciated.
Hello,
I'm trying to verify secure boot on LAVA lab, I've verified the positive case of secure boot and confirmed the secure boot is enabled from the boot logs (kernel-start-message = "EFI stub: UEFI Secure Boot is enabled.").
Next, I would like to verify the negative test case as well by tampering the secure boot image where I expect the image boot shall fail.
Is there way in LAVA to make the job complete(e.g. waiting for specific message in boot logs like "Can't find image information. ") even though the image fails to boot on the DUT?
Ref. https://lava.ciplatform.org/scheduler/job/1236667#L81
Hi,
I am not able to submit multiple jobs on my single node lava server, currently even 1 job is running the second job waits till the 1st job is completed. I have tried to increase the job limit of my worker to 3 via lava Admin UI but still I cannot run 2 jobs in parallel. Could you please let me know how can I increase the no. jobs.
Hi,
I have been looking into how to secure LAVA, as so far it's just been learning how it works. There's the obvious http/https stuff, and in the documentation mentions firewalls for remote workers (which we'll be using anyways). We already have things in place for physical security and plans on how to improve, so no worries there.
Is there anything not covered in the documentation that should be considered and how could we go about putting that into place?
Best regards,
Michael
I saw a file 'lava-dispatcher/actions/boot/gdb.py' in the LAVA source code, which seems to define a gdb method in action->boot. But I didn't find a description of it in the LAVA documentation. The document I checked is `https://docs.lavasoftware.org/lava/actions-boot.html?highlight=boot#boot-action-reference`.
Reading the source code may lead to some misunderstandings and is time-consuming. I hope someone who knows its function can explain to me what this new method does. Thank you very much!
I want to add some key:value pair in metadata of json that POST by notification of job, but I can not figure it out. I had tried add metadata in job defination and test defination, but the back json did not have my data in metadata field, it was just a [].
Then I read the lava source code, I find the metadata data are produced by method `get_metadata_dict()` in class `TestJob`. This method is called at method `create_job_data` in class `TestJob`, it creates the json data which will post to callback url. `get_metadata_dict()` check whether there is an attribute `testdata`, but I can not find the place where this attribute‘s assignment.
Could someone give me a way to do this? Thanks you very much! 🙏
Hello,
We are testing our custom debian image and have added banner messages before login and after login and after the implementation LAVA job is failing on auto-login.
Please find example banner messages:
|_ _/
| |
| |
_| |
|_____\
Info:
Example.....: ABC
Example......: ABC
IP:
etho.......:123
Info:
1. some info
Some Info:
1. ###
Some more info: ####
login :
It looks like due to a huge banner message before the login prompt it is getting confused and due to this, it gets stuck at auto login and timeouts. Please suggest if we can ignore this in LAVA to login and execute the test case
Please find the job details:
device_type: qemu
job_name: banner
timeouts:
job:
minutes: 30
action:
minutes: 25
connection:
minutes: 5
priority: medium
visibility: public
context:
arch: amd64
lava_test_results_dir: "/home/lava-%s"
actions:
- deploy:
timeout:
minutes: 15
to: tmpfs
images:
rootfs:
url: file:///home/image
image_arg: XXXX
- boot:
timeout:
minutes: 15
method: qemu
media: tmpfs
prompts:
- 'login:'
auto_login:
login_prompt: "login:"
username: abc
password_prompt: "Password:"
password: "abc"
- test:
timeout:
minutes: 15
definitions:
- repository: git(a)test.git
from: git
path: test.yaml
name: basic
Thanks,
Sweta
Hello,
I'm currently using version 2024.09, and when running the test monitor, I encounter the following error at the top of my job: ('Connection aborted.', RemoteDisconnected('Remote end closed connection without response')). I don't notice anything unusual in the test logs. Could this be a bug with the test monitor?
Best Regards,
Andy
Hi,
I think there is something terribly wrong with job timeouts in 2024.05
release. Total job timeout seems to be ignored. This was working fine
on 2024.01 release. I'll try to debug, but IMHO this is a pretty
critical bug that should be addressed immediately. I'll report it in
gitlab as well.
Best Regards,
Milosz
Hello,
I am spinning up debian image using Qemu and want to copy a custom package from my lava server to debian vm for testing. I could see that support for tar is planned https://docs.lavasoftware.org/lava/actions-test.html#inline-test-definition… . But is there any way I can include my custom package in tarbar created at Deploy Action to ship to my VM easily?