#Hello!
1 messages Β· Page 1 of 1 (latest)
It should just work β’?
You can use the docker client on macOS after colima start with no additional setup.
When you call dagger call it will also by default find the docker client which I think would point to colima.
Oh interesting, just saw Marks comment π
Excited to learn more about your specific use case.
Ran some linters with/without colima.
Numbers:
- With Colima: 7m
- Without Colima: 1m 40s
Could be any number of reasons. I recall qemu io being slow a couple years back. Dunno if that's been fixed.
Thanks @gusty star - feels like may be related to some issues we've seen with things like podman too - ill make an issue.
If podman uses qemu under the hood too, it might be related, yes.
Could be any number of reasons. I recall qemu io being slow a couple years back. Dunno if that's been fixed.
is colima using qemu though? there's something strange here... shouldn't it should spin a native aarch64 vm?
seems like qemu + hvf should be ok. I don't have a mac machine to try π¦
@soft elbow happy to assist if you give it a try π
Hey @fathom latch . I will definitely give it a try. Our team (release engineering) has the ability to use docker desktop, but our developers may not be so lucky.
I will set up colima and try to use it for a simple example on both, see if that just worksβ’
We are evaluating Dagger as I mentioned and we had a couple of questions that I'll pop over here if you don't mind giving us some directions π
- Accessing private repositories (e.g. go private repository dependencies referenced on a dockerfile) / Accessing private docker registries
- Is "build me this dockerfile" the standard approach when building containers? Is translating a dockerfile to dagger code the recommended approach when building containers?
- Stable release of dagger. Is there any visibility on the roadmap? Any plans on when that could be a thing?
I will be more hands on next week with more technical questions and hands on those questions, thanks for the replies before hand!
Hey @soft elbow thanks for the great questions heres my response.
Accessing private repositories (e.g. go private repository dependencies referenced on a dockerfile) / Accessing private docker registries
Dagger has first class support for Secrets, you can learn more about it here: https://docs.dagger.io/manuals/developer/secrets
For working with private containers, the Container type has support for authenticating in a first class way: https://docs.dagger.io/api/reference/#Container-withRegistryAuth
Is "build me this dockerfile" the standard approach when building containers? Is translating a dockerfile to dagger code the recommended approach when building containers?
It depends on many factors, my rule of thumb is if you've got an existing dockerfile you can pass that along. If you're starting from scratch you'll get some additional benefits by writing it in dagger.
Dagger supports reading Dockerfile. We're still working on porting this part of the docs over to the new docs page but the general idea should be the same as shown here: https://archive.docs.dagger.io/0.9/cookbook#build-image-from-dockerfile
Stable release of dagger. Is there any visibility on the roadmap? Any plans on when that could be a thing?
We are currently working toward a 1.0 release of Dagger. We think the engine is in a pretty stable state and right now we are working on a lot of DX polish. We do not intend to drastically break the API in the near future.
Dagger allows you to utilize confidential information, such as passwords, API keys, SSH keys and so on, in your Dagger Modules and Dagger Functions, without exposing those secrets in plaintext logs, writing them into the filesystem of containers you're building, or inserting them into the cache.
Filesystem
Thank you very much for the responses @strong sorrel , amazing. We'll be running a few spikes over the next week(s) to port many of our pipelines into dagger. I'm sure I'll be back here soon π
Heya @gusty star ! We are evaluating dagger and running some tests with Docker Desktop and Colima. Any chance you could let me know what you ran that produced these performance differences? So far I am seeing much closer numbers; Colima generally running more slowly but not this egregious.
Hey Robert!
I just ran the lint function in one of my projects: https://github.com/openmeterio/openmeter
The numbers I got could be the result of any number of reasons. π