linaro-validation

linaro-validation@lists.linaro.org

818 discussions

by Andy Doan

I noticed today while looking at a merge proposal that we store some test data files for LAVA on the wiki[1]. I'm sure that was chosen because it works and its easy to manage. However, I'm not a huge fan of this for a few reasons: * easy to delete an attachment * Moin provides no revision control over attachments * I took this job to get out of the wiki business :) I was thinking of two alternate ways to manage our data files: 1. Create something on people.linaro.org like the toolchain team did[2]. 2. Create something like validation.linaro.org/testdata Seems like option 1 is the easiest and has been done before. Any opinions? 1: https://wiki.linaro.org/TestDataLinkPage 2: http://people.linaro.org/~toolchain/

14 years

Daily Graphics WG test jobs

by Alexandros Frantzis

Hi all, the Graphics WG needs to run daily graphics-oriented test jobs (actually, we only need to run them whenever something changes, but I think this is not implemented yet). We had previously scheduled a daily job as a cron job in the lab, but it seems that that got lost at some point. Submitting test jobs from a remote machine (e.g. my desktop) is not really an option, since I think it's important for job runs to be independent of the availability of any one person's computer and network. Our needs have begun to grow (we need more test jobs) and I would also like to experiment a bit, so pestering someone from the validation group to add or modify scheduled runs doesn't seem very efficient. I would much prefer to have an official mechanism that didn't require the direct intervention of the validation group (e.g. "My scheduled jobs" in the v.l.o web site). We would like to have a set of test jobs running again for 12.05. How do you propose we should handle this in the short-term? Thanks, Alexandros

14 years

token management for automatically submitted jobs

by Michael Hudson-Doyle

Hi all, Suppose there is a LAVA user, and to avoid taxing my imagination let's call him Alexandros. He wants to have some jobs submitted automatically from ci.linaro.org to lava that deposit results in a bundle stream that only members of linaro can see, which all seems reasonable enough. Currently though, the story for tokens around this is a bit horrible. To be able to submit to the a /private/team/linaro/... bundle, you have to submit the job as a member of the linaro group in v.l.o. I can think of a few ways of doing this, but I don't really like any of them: 1) jenkins on ci.linaro.org could use one of alf's tokens, but that seems a little tied to him (what if he leaves linaro, etc) 2) Another way is to create a user that does not correspond to a user on LP (gfx-daily-job-submitter or somethign) and add it to the linaro group on v.l.o. This feels a bit better, but it's not very 'self service' -- the only way to create such a user is via the admin panel afaik. 3) A third way is to create a fake user on LP and add it to the ~linaro team there. This also seems a bit horrible. There is a fourth way that is actually happening but doesn't help -- create a user on LP and do _not_ add it ~linaro: https://launchpad.net/~ciadmin [1]. I don't really have a suggestion for what would be better here. It feels a bit like the model we have for access and handling tokens is perhaps a bit too simple currently. What do you guys think? Cheers, mwh [1] this is why ci.linaro.org lost the job-submitting permission -- I didn't realize ciadmin on v.l.o corresponded to a user on LP!

14 years

#linaro-lava

by Zygmunt Krynicki

Hi folks. I'd like to propose that we keep all the lava discussion in #linaro-lava if possible, this will allow participants, who are not working for linaro, to join and quickly identify people that share the interest in our common framework. The channel name is a compromise between unavailable #lava (already owned by unrelated project) and staying in #linaro (that sees a fair amount of unrelated traffic). I also considered #linaro-validation but I think we have agreed in the past that we want to transition away from the "validation" keyword to "lava". This is not formally done yet, I'd like to have a chanserv registry (I think all #linaro-* channels are automatically managed though) and public logs, much like we have on #linaro today. So, if you support this idea and find yourself talking or observing a discussion about LAVA in #linaro please gently suggest the participants to move to #linaro-lava Thanks Zygmunt Krynicki

14 years

the need for three instances: production, staging and ... dogfood?

by Michael Hudson-Doyle

Hi gang, As you may have seen if you actually read all the merge proposal email you get <wink>, as part of my quest to get us closer to a continuous delivery stype of development I've written a script that updates a LAVA instance to the tip of trunk of all the various LAVA components. I will set things up so that this runs every night on the staging instance soon (need to fiddle things somehow so that a sudo password isn't necessary to restart the instance after the code is upgraded). However, Spring is right now using the staging instance to test some dispatcher changes. If I'd set the cronjob I refer to above up and Spring was particularly unlucky, the cronjob would wipe out the changes he's made to the instance to test while he's testing them. This seems less than ideal. Because what we do is always going to be hard to test outside a production like environment, I think what we need is yet another instance: one that we can hack about relentlessly as needed to test changes that we haven't yet landed. I propose the name "dogfood" for this (maybe we'd even go as far as having scripts that completely removed the dogfood instance and could recreate it as needed to ensure a clean baseline). So we'd have: * production: duh * staging: this is _solely_ used to determine if changes that have landed to the trunk of some component are safe to deploy to production * dogfood: random hacking I think that potentially we could have some way of bringing up dogfood instances as needed in our private cloud, but tbh for now one on control that we share amongst ourselves in a loose way (i.e. check on IRC before doing stuff with it) would be a useful thing. What do you guys think? Cheers, mwh

14 years

LAVA trunks _should_ work with Django 1.4 now

by Michael Hudson-Doyle

Hi all, I've made a few small changes to lava-server and linaro-django-xmlrpc to work with Django 1.4. I have clicked around a bit locally, but certainly not hit each page, so if you can upgrade your local Djangos and let me know how it goes, that'd be great. Cheers, mwh

14 years

test ran, results didn't show

by Andy Doan

I've come across an error I haven't seen before: <http://validation.linaro.org/lava-server/dashboard/streams/public/personal/…> The job log is: validation.linaro.org/lava-server/scheduler/job/17017/log_file If you look at the bundle viewer, you'll see more tests were run but we didn't gather their results. Can anyone explain what's happening here?

14 years

identifying USB devices

by Michael Hudson-Doyle

Part of the whole SD card mux game involves finding the card reader the SD card for a particular board is plugged into. Following a lead provided by Zygmunt, it seems that you can address devices by USB topology by looking in /sys/bus/usb/devices/ -- for example, the front right USB port on my laptop seems to correspond to a directories called "1-1.2" and "1-1.2:1.0" in here, the back left port corresponds to "3-1" and "3-1:1.0" and a particular port on a USB hub plugged into the front left USB port seems to correspond to "2-1.2.3" and "2-1.2.3:1.0". So this seems to be reasonably straightforward (although I don't know if the mapping is necessarily stable across reboots or kernel upgrades -- seems like it should be though). The next fun is mapping this directory to a block device. Poking finds that (with the card reader plugged into the last location mentioned above) the directory at: /sys/bus/usb/devices/2-1.2.3:1.0/host31/target31:0:0/31:0:0:0/block/ contains a directory called 'sdb' and indeed the sd card is mounted as /dev/sdb. Playing around shows that the 31 here is a number that gets incremented each time you plug/unplug the reader (or maybe any USB device). So this suggests that we could address the card reader in this port as "2-1.2.3:1.0" and to e.g. run l-m-c targetting the card in the reader, we should look at /dev/$(ls /sys/bus/usb/devices/${address}/host*/target*/*:0:0:0/block) which seems like it would work but frankly also seems like a total hack. Do any of you know if this can be done in a cleaner way? Cheers, mwh

14 years, 1 month

RFC Patch: lava with no media files

by Andy Doan

I was trying to get my local LAVA instance to work with our lab's data base dump. So I had the DB, but not all the media files. LAVA almost works when you do this, but I hit two small issues that prevented me from looking at bundle streams and it their test runs. I realize this is an edge case, but its handy for when you want to test with a real database, but don't want all the media files. I'm not sure this is the exact way we want to fix this, so rather than do a merge proposal I thought I'd get your thoughts on the patch first. There were two main issues in the code: bundle_detail wants to get the "document format" for the media file associated with the bundle. Since the file didn't exist we hit an exception. I added a small check to the get_document_format to handle this by returning "n/a". bundle_detail and test_run_detail wanted to display the size of media files. This calls a new function content_size which will catch an error if the file doesn't exist and simply return 0. I'm not sure if this patch is doing to good a job of hiding the fact an error does exist. Maybe we should put a conditional in the template file to display an error when the file doesn't exist also? -andy

14 years, 1 month

master boot image and uboot commands

by Andy Doan

I've been setting up a local LAVA lab on my home network. I've used the lava-deployment-tool and lava-master-image-scripts to get my Beagle hooked up. I'm really close to having everything work. However, when I submit a job I hit a problem with the soft-reboot logic not being able to instruct u-boot to use the test image. I'm flashing a 12.01 pre-built nano image, but I don't think that matters. I think the real issue might be something with a newer uboot being used by the linaro-master-image-script and the device.conf for beagle not having the right stuff. Has anyone else run into this or have suggestions? I tried briefly updating the .conf file for beagle, but it didn't seem to stick. It almost looked like we were sending the commands before u-boot was ready, but that may just be how the output is displayed. -andy

14 years, 1 month

← Newer
1
...
70
71
72
73
74
75
76
...
82
Older →

Jump to page:

2026

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

linaro-validation