| Age | Commit message (Collapse) | Author |
|---|
| |
| [r=bladernr][bug=][author=rodsmith]" |
| |
| The usb3 superspeed performance test is using the exactly same command with usb3 storage test, there is no need to run it twice. |
| The pulse-active-port-change script was using pairs (sink-source-name, sink-source-active-port) to detect changes when plugging a microphone or a headset. This was not enough because in some cases (on desktops especially) the pair would remain the same, which would fail the test although the feature actually worked. By adding the availability flag, we make sure any mic/headphone plugged will be detected properly. |
| |
| [r=zyga][bug=1274657][author=cypressyew]" |
| --cpu_load_fail_threshold |
| and changed to dashes in --cpu-load-fail-threshold, as per Jeff's suggestions |
| add multi_iperf3_nic test to use these new features |
| It looks like X crashes when trying to re-enable Unity Shell Compiz plugin, after the QML touch test is run. In order to avoid that, we add a sleep command just before re-enabling Unity Shell Compiz plugin. |
| Remove the outdated obex_send script, which has already been replaced by bluetooth_test script, also remove those corresponding test cases. |
| This patch adds category ids to all jobs in the checkbox provider (almost all jobs) so that users can select the category they want to test in checkbox-converged. local jobs are not getting a category id as those will be run during bootstrapping. Legacy jobs in install, panel-reboot, panel-control and server-services are also kept uncategorised. |
| [r=zyga][bug=1471945][author=cypressyew]" |
| Add an .gz extension to those fwts attachment jobs that runs with gzip command, make user know what kind of attachment they're decoding. |
| |
| |
| Move the DEBUG information to after the ERROR message, make it less confusing for the tester. |
| [r=sylvain-pineau,bladernr][bug=1486820][author=cypressyew]" |
| [r=bladernr][bug=][author=rodsmith]" |
| |
| alternative to client or server test type |
| Divide the CERT tests into QA and HWE tests, so we don't need to split the log in fwts_test, we can create different logs with different jobs. |
| Add the FWTS HWE attachment job for server provider, as it's using a different log filename |
| |
| fwts_test called without any parameter will launch all the certification tests. In that context, all fwts test outputs will be grouped into a single log file. This is where we need to split hwe tests into a second log file. But there's also a local jobs creating jobs for all tests returned by fwts_test --list. Here not only cert tests will be returned but all the tests defined in the script (aka advanced tests). This execution mode is meant to be used by CDTS where we wanted to have a job per fwts test and thus a log per test. |
| ethernet/multi_nic that are no longer used |
| Add the post-suspend clicking test. |
| |
| |
| engineering-sample status without the other checks, and created a new job definition to use this test. |
| tarmac [r=zyga][bug=1483829][author=bladernr]" |
| |
| |
| |
| |
| |
| |
| |
| |
| on detected disks |
| |
| tarmac [r=zyga][bug=1485451][author=cypressyew]" |
| Add post-suspend external audio test cases. |
| Add the post suspend oops test and the corresponding log job. |
| 'smartctl -i' but then says it doesn't support SMART logging via 'smartctl -l'. |
| |
| This patch changes the following attachment jobs: firmware/fwts_desktop_diagnosis_results.log firmware/fwts_desktop_diagnosis_results_hwe.log installer_bootchart.tgz power-management/poweroff-log-attach power-management/reboot-log-attach stress/graphics-tarball stress/poweroff_30_check_log stress/poweroff_30_log stress/poweroff_check_log stress/poweroff_log stress/reboot_30_check_log stress/reboot_30_log stress/reboot_check_log stress/reboot_log suspend/xrandr_screens_after_suspend.tar.gz In addition, the following local job, which generates attachment jobs, is also changed: suspend/xrandr_screens_after_suspend.tar.gz There are several changes applied here. First, for jobs that use this, the test condition is changed from -e to -f. The rest of the code won't cope with non-file input so it's better to be explicit. While it might, perhaps, be better for those jobs to fail, the condition is there because there is no explicit dependency (because we want to attach log files even if, or especially when, the related non-attachment job fails. The most important change is the way those two attachment jobs operate. Instead of compressing the file and then dumping the binary, compressed data each of the log files is compressed on the fly (to stdout) and base64 encoded. This makes the job re-runnable (as the original log file is no longer removed by gzip) and, most importantly, the output is always valid ASCII text. This should address the issue where those attachments would just end up being empty in the submission. As a small detail, to ensure that the errors are propagated a bash-specific 'pipefail' option is enabled. This lets simple constructs such as: tar cvz /path/to/some/things | base64 Behave correctly when tar fails for any reason. This change was only made to jobs that don't already use the [ -f ... ] && construct as it seems not to be needed there. Fixes: https://bugs.launchpad.net/plainbox-provider-checkbox/+bug/1479648 Signed-off-by: Zygmunt Krynicki <zygmunt.krynicki@canonical.com> |
| tarmac [r=zyga][bug=1481176][author=cypressyew]" |
| [r=zyga][bug=][author=kissiel]" |
