lava-users

lava-users@lists.lavasoftware.org

4 participants
901 discussions

bootup a server and deploy debian bookworm over serial connection

by sweta.ghosh＠nagarro.com

Hello, I am new to LAVA, I want to boot up my hardware(NET-I System)which is a x86_64 and install debian bookworm image over it and execute test cases. My h/w is connected to my LAVA server via a serial device.Below is the high level job that I am planning to use. But where should I specify the serial device information and other information?Could please share any example job and device .jija2 that I can refer regarding what all information needs to be provided for LAVA server to properly connect to H/W and also is u-boot correct for my usecase? or if there is any other suggestions : device_type: your_device_type job_name: debian_boot_test timeouts: job: minutes: 15 action: minutes: 5 connection: minutes: 2 actions: - deploy: timeout: minutes: 10 to: tftp images: rootfs: url: <Debian Image> - boot: method: u-boot commands: your_device_boot_commands prompts: - 'login:' parameters: boot_options: root=/dev/ram0 - test: timeout: minutes: 5 definitions: - repository: metadata: format: Lava-Test Test Definition 1.0 name: debian-check-test description: "Test to verify Debian OS" os: - debian run: steps: - echo 'Checking if OS is Debian' - 'if [ "$(lsb_release -is)" = "Debian" ]; then echo "OS is Debian"; else echo "OS is not Debian"; fi' from: inline name: debian-check-test path: inline/debian-check-test.yaml

3 days, 21 hours

Change path of lava_test_results_dir and lava_test_dir

by sweta.ghosh＠nagarro.com

Hello, I would like to change path to /mnt from below: "lava_test_dir": "/lava-%s", "lava_test_results_dir": "/lava-%s", I am using an immutable image and due to which don't have permission to write, so cannot create directory eg /lava-34 due to which cannot execute test. We have partition /mnt which we can mount and run the test cases over our os. But for that to work I need to change path. I tried changing it in /usr/lib/python3/dist-packages/lava_dispatcher/deployment_data.py , but jobs fails once I change it. I have only changed the debian section as the image we are using is a debian image. Could you please help me if there is any option to make these changes while executing job or any other file for devices. Thanks, Sweta

2 weeks, 3 days

RecursionError: maximum recursion depth exceeded on lava-scheduler service

by sweta.ghosh＠nagarro.com

Hi , I am trying to setup LAVA on Single Master Worker node. When i am trying to submit job it is running but no device is getting selected nor I am getting any output on UI. When I checked the lava-scheduler it is failing with below error: root@debian:/etc/apache2/sites-available# systemctl status lava-scheduler × lava-scheduler.service - LAVA scheduler Loaded: loaded (/lib/systemd/system/lava-scheduler.service; enabled; preset: enabled) Active: failed (Result: start-limit-hit) since Tue 2024-07-02 18:52:36 IST; 7min ago Duration: 1.472s Process: 10345 ExecStart=/usr/bin/lava-server manage lava-scheduler --level $LOGLEVEL --log-file $LOGFILE $EVENT_URL $IPV6 (code=exited, status=0/SUCCES> Main PID: 10345 (code=exited, status=0/SUCCESS) CPU: 1.431s Jul 02 18:52:35 debian systemd[1]: lava-scheduler.service: Deactivated successfully. Jul 02 18:52:35 debian systemd[1]: lava-scheduler.service: Consumed 1.431s CPU time. Jul 02 18:52:36 debian systemd[1]: lava-scheduler.service: Scheduled restart job, restart counter is at 5. Jul 02 18:52:36 debian systemd[1]: Stopped lava-scheduler.service - LAVA scheduler. Jul 02 18:52:36 debian systemd[1]: lava-scheduler.service: Consumed 1.431s CPU time. Jul 02 18:52:36 debian systemd[1]: lava-scheduler.service: Start request repeated too quickly. Jul 02 18:52:36 debian systemd[1]: lava-scheduler.service: Failed with result 'start-limit-hit'. Jul 02 18:52:36 debian systemd[1]: Failed to start lava-scheduler.service - LAVA scheduler. root@debian:/etc/apache2/sites-available# tail -f /var/log/lava-server/lava-scheduler.log {% extends 'qemu.jinja2' %} ^^^^^^^^^^^^^^^^^^^^^^^^^ [Previous line repeated 977 more times] File "/usr/lib/python3/dist-packages/jinja2/environment.py", line 1494, in is_up_to_date return self._uptodate() ^^^^^^^^^^^^^^^^ File "/usr/lib/python3/dist-packages/jinja2/loaders.py", line 212, in uptodate return os.path.getmtime(filename) == mtime ^^^^^^^^^^^^^^^^^^^^^^^^^^ RecursionError: maximum recursion depth exceeded Also when I am accessing /admin/lava_scheduler_app/device/ I am getting " 500 Internal Server Error maximum recursion depth exceeded" I have registered my worker node as debian,hostname as debian and device type is qemu. Could you please what I am missing.

2 weeks, 4 days

lava test doesn't wait for command to finish running

by Michael Peddie

Just as the subject says. I am using lava-test-case to confirm whether this particular command is done running and is successful or not, because the test has previously always ended prematurely and not done what I needed it to do. When I run the command manually outside of a lava test job it works just fine, no errors and doesn't end early. It does take a while to complete, but I don't think that's the issue as I use wget to download some large files and those have taken several minutes longer than this one command is supposed to. How do I get lava to wait for this command to end? Or to change how it checks for command failure? I get "Received signal: <ENDTC>" almost immediately after running the command, and subsequently the result=fail signal. I really don't know why it won't work, so if there is a change I can make to the files or a config somewhere I would like to know. Best regards, Michael

3 weeks, 5 days

MultiNode to run script->reboot->test

by Michael Peddie

Hi, I was wondering if it is possible to write a multinode test to boot a device, run a script to do some stuff then have it reboot one way or another, then run tests. There is intention for pipeline integration down the line, so if it isn't possible to do that with multinode I have considered the idea of setting it up so one regular test runs and runs the script, turns DUT off, then another test runs to boot and actually test the device. This (obviously) has its drawbacks, but it's something of a backup. If I understand correctly, where certain actions are placed in a multinode test actually matters, so it (in my mind) is possible to have a boot->test to boot a DUT then use test commands to run a script, and then after that another boot->test to boot the device or just login again and then actually run tests. So the final sort of action set up would be boot->test->boot->test, where the first two are for running the script and rebooting, and the latter two are actually testing. If there are other ideas that are better or equally viable, then Im open to those too, but multinode was the first that came to mind cos of secondary connections and the like. Best regards, Michael

2 months, 1 week

lavasoftware.org is unaccessible from gandi.net Paris SD6 datacenter

by Jacopo Mondi

Hello I am experiencing troubles connecting to apt.lavasoftware.org from a VPS located in the "Paris France SD6" datacenter hosted by gandi.net, The problematic VPS IP address is 46.226.105.174. Testing from another VPS, still hosted at gandi.net but located in a different datacenter (Luxembourg) presents no issues. Both VPS runs Debian 12 Bookworm. The symptoms I'm experiencing seems to suggest the TLS handshake gets truncated as suggested by the 'gnutls-cli' output below reported ------------------------------------------------------------------------------- # gnutls-cli -d9999 -p 443 apt.lavasoftware.org ... |<5>| REC[0x56498f5d7740]: Sent Packet[1] Handshake(22) in epoch 0 and length: 402 |<11>| HWRITE: wrote 1 bytes, 0 bytes left. |<11>| WRITE FLUSH: 402 bytes in buffer. |<11>| WRITE: wrote 402 bytes, 0 bytes left. |<3>| ASSERT: ../../lib/buffers.c[get_last_packet]:1185 |<10>| READ: Got 0 bytes from 0x3 |<10>| READ: read 0 bytes from 0x3 |<3>| ASSERT: ../../lib/buffers.c[_gnutls_io_read_buffered]:593 |<3>| ASSERT: ../../lib/record.c[recv_headers]:1195 |<3>| ASSERT: ../../lib/record.c[_gnutls_recv_in_buffers]:1321 |<3>| ASSERT: ../../lib/buffers.c[_gnutls_handshake_io_recv_int]:1467 |<3>| ASSERT: ../../lib/handshake.c[_gnutls_recv_handshake]:1600 |<3>| ASSERT: ../../lib/handshake.c[handshake_client]:3075 |<13>| BUF[HSK]: Emptied buffer *** Fatal error: The TLS connection was non-properly terminated. ------------------------------------------------------------------------------- Could you kindly check if my VPS is maybe in a range of IP addresses blacklisted by the service provider that hosts lavasoftware.org ? Thanks j

2 months, 1 week

Boot with serial test with ssh

by Michael Peddie

Hi all, I have been having issues doing solely boot and test with serial (specific issue(s) not diagnosed, but presumably something to do with telnet), but that's not the discussion here. I want to be try booting the device with the serial method then deploying/running tests using ssh, since it seems like ssh can't power on and power off the DUT. If that is wrong then please let me know so I can try that. I have two proposed ideas: 1) boot with serial -> deploy test (however works) -> run test with serial -> power off 2) boot with serial -> deploy test (however works) -> run test with ssh -> power off I started with a multinode job (I still don't quite understand multinode yet) and this is my current job definition, if you could please help point out issues with it as it doesn't work: #multinode job for controller deployment job_name: controller deploy test protocols: lava-multinode: roles: host: context: lava_test_results_dir: /tmp/lava-%s device_type: controller timeout: minutes: 10 count: 1 guest: context: lava_test_results_dir: /tmp/lava-%s request: lava-start count: 3 expect_role: host timeout: minutes: 10 connection: ssh host_role: host timeouts: job: minutes: 15 action: minutes: 5 connection: minutes: 2 priority: medium visibility: public actions: - deploy: role: - host timeout: minutes: 10 to: tftp authorize: ssh kernel: url: file:///kernel.img type: uimage ramdisk: url: file:///ramdisk.gz compression: gz dtb: url: file:///u-boot.dtb - deploy: role: - guest timeout: minutes: 10 to: ssh protocols: lava-multinode: - action: prepare-scp-overlay request: lava-wait messageID: ipv4 message: ipaddr: $ipaddr - boot: role: - host timeout: minutes: 5 method: minimal prompts: ["# "] auto_login: login_prompt: "login: " username: root - boot: role: - guest timeout: minutes: 5 prompts: ["# $"] parameters: hostID: ipv4 # messageID host_key: ipaddr # message key method: ssh connection: ssh - test: role: - host timeout: minutes: 15 definitions: - repository: metadata: format: Lava-Test Test Definition 1.0 name: install-ssh description: "install step" scope: - functional run: steps: # messageID matches, message_key as the key. - lava-send ipv4 ipaddr=$(lava-echo-ipv4 eth0) - lava-send lava_start - lava-sync clients from: inline name: ssh-inline path: inline/ssh-install.yaml - test: role: - guest timeout: minutes: 5 definitions: - repository: https://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@github.com/MichaelPed/lava… from: git path: smoke-tests/smoke.yaml name: smoke-tests # run the inline last as the host is waiting for this final sync. - repository: metadata: format: Lava-Test Test Definition 1.0 name: client-ssh description: "client complete" scope: - functional run: steps: - df -h - free - lava-sync clients from: inline name: ssh-client path: inline/ssh-client.yaml And this is my device dictionary, if you could please let me know what needs changing: {% extends 'controller.jinja2' %} {% set power_off_command = 'python /power.py 12 off' %} {% set soft_reboot_command = 'reboot'%} {% set hard_reset_command = 'python /power.py 12 reboot' %} {% set power_on_command = 'python /power.py 12 on' %} {% set connection_list = ['uart0'] %} {% set connection_commands = {'uart0': 'telnet 10.60.2.209 7001'} %} {% set connection_tags = {'uart0': ['primary', 'telnet']} %} {% set bootm_kernel_addr = '/kernel.img' %} If you guys have any questions please ask away! Also, if there is somehow something other than a multinode job that will do what I want then please let me know! Best regards, Michael

3 months, 1 week

How to run QEMU in DUT and gdb the QEMU in another console

by 叶轩杨

Hello! I want to run QEMU in DUT and gdb the QEMU in another console to simulate something like a bit flip in memory. What should I do? I just need the new console to run some gdb scripts, like `rust-gdb vmlinux -ex 'target remote /tmp/gdb_socket' -ex 'set *0x40001000 ^= (1<<5)' -ex `c` `, so the new console can not be interactive, live to run some command is enough. Thanks!

3 months, 2 weeks

Run pytest suite on DUT using LAVA

by Singamsetty Nagendra

Hello everyone, I would like to integrate & run a bare metal custom tests suite(pytest) on lava. Currently I am executing manually from Host Machine(where pytest is configured and pytest suite will execute) and these commands are sent to DUT and results are collected on Host Machine. I am familiar with lava-interactive method but it looks like it will not suit this requirement(Run pytest on Host and collect logs from DUT). Does lava suits for this requirement? if someone could give me advice on this that would be great support. Regards Nagendra S

3 months, 2 weeks

How to loop test reboot

by irreallich＠126.com

Hi, I hope to be able to test whether the system can boot normally after multiple restarts (10,000 times) Because every boot requires the auto_login of the boot action, it seems that a loop needs to be implemented in yaml, and this loop needs to include the boot action. Are there any relevant examples that you can refer to? There are roughly two types of test logic 1: while (a<10000) { -boot auto_login - test basic io test software reboot in target board a=a+1 } 2: while (a<10000) { -boot auto_login - test basic io test a=a+1 hardware reboot in worker } Thanks

3 months, 2 weeks

Jump to page:

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

lava-users