Closed ar-jan closed 2 years ago
Looks like we can also use a predefined Xvfb Action from the marketplace for running the headless test.
Prefer not to add further dependencies, and looks like it shouldn´t be needed for this.
Test fails, perhaps because it keeps using Chrome rather than ChromeHeadless?
Launching browsers Chrome with concurrency unlimited Starting browser Chrome Cannot start Chrome ERROR:file_io_posix.cc(144)] open /sys/devices/system/cpu/cpu0/cpufreq/scaling_cur_freq: No such file or directory (2) ERROR:file_io_posix.cc(144)] open /sys/devices/system/cpu/cpu0/cpufreq/scaling_max_freq: No such file or directory (2)
On IAnalyzer the log for this step reads Launching browsers **ChromeHeadless** with concurrency unlimited
.
I've matched the configuration for karma.conf.js with that in IAnalyzer (and als tried adding the --disable-gpu
and --no-sandbox
flags which were used on Travis and are often mentioned as needed for running headless Chrome), but the test still tries to start Chrome. I've also tried base: 'ChromeHeadless'
instead of Chrome
in the customLauncher, same result.
Any ideas, @oktaal @jgonggrijp?
Not really, unfortunately. Haven't seen this problem before. Maybe @oktaal has? Otherwise, either of us could probably study this a little bit next week.
Otherwise, either of us could probably study this a little bit next week.
That would be great!
I forgot about this, sorry about that. Fortunately, it looks like @oktaal is working on it.
karma.conf.js
was in the wrong location. Now it works!
Great!
Ready to merge?
- Handling for Firefox and Chrome?
- [ ] Cache directory declarations?
- [ ] Divide frontend, backend, and functional tests in three steps?
Sure! Might need to move these things to separate issues then.
Merged. Opened followups #29 and #30. I left out cache dirs (was based on Travis), I suppose we'll find out if/when we need that.
You never really "need" cache dirs. They are a way to save CPU cycles, hence CO2 emissions.
More or less a port of the .travis.yml contents, may have to be adapted further (#25).
Modify backend installation to allow using a PostgreSQL connection string with TCP port and passwordProviding PostgreSQL environment variables with TCP port is sufficient