The new CACHED status in the TUI for Dagger 0.15.0 is really useful. However the logs returned from a Dagger call in my CI systems (Gitlab/GitHub) don't show the same sort of information, and don't have the CACHED keyword. I guess because the TUI isn't used in this instance for the CI logs. Is there a simple way to identify when commands are using the CACHE in CI logs?
#Identifying CACHED in CI Logs (Gitlab/GitHub)
1 messages · Page 1 of 1 (latest)
Hey @finite latch can you tell us more about how you are running the dagger engine in CI?
I suspect the problem is that nothing is cached in CI because the default approach is to run the engine on ephemeral runners.
Hey @finite latch I did some experiments and figured out that the core issue is that the cached indicator only shows up when you are running dagger inside of a TTY.
I experimented with faking a tty in gitlab but didn't get very far, the output looks even worse than in plain mode. https://gitlab.com/levlaz/snippetbox/-/jobs/8802900495
I think the simplest way to see this info is to use Dagger Cloud to look at the CI logs.
Otherwise, I don't think there is a good option. Please feel free to add a feature request in Github if you'd like to see the cached indicator added to plain output.
GitHub
Signed-off-by: Marcos Lilljedahl marcosnils@gmail.com
Hi, ah yes, I keep forgetting that ephemeral runners are also used. I should've mentioned these are on self-hosted runners and I can kind of work out they are using cached results from the very low runs times (e.g. 0.0s)
Thanks for the investigation.
And wow you guys work quickly, so that feature should appear in a future version of Dagger?
yup, it should land in v0.15.2 - the idea is to release that early next week 🎉
You can use it right now if you install dagger from the main commit as well
If you're installing Dagger with the install script here's an example on how to install it from a specific commit #maintainers message
@hollow abyss you are too good haha 🙂