These instructions were not up-to-date (they didn't mention
the different storage backends, and some variables were
tagged as optional while they were mandatory). With this
update, they should (hopefully) be more accurate! :)
I also added instructions if someone wants to run Nixery
outside of the container image (I found it convenient when
working on Nixery's code).
This is required by common patterns in shell scripts.
There are some caveats around this. Adding logic to filter whether
coreutils is included in an image would slow down the Nix evaluation,
so the link is currently created even in cases where it doesn't point
to anything.
Fixes#109
After the discussion in #116, this stores the blob content types
in extended attributes when using the filesystem backend.
If the underlying filesystem doesn't support extended attributes,
storing blobs won't work; also, if extended attributes get removed,
blobs won't be served anymore. We can relax this behavior if
needed (i.e. log errors but still accept to store or serve blobs).
However, since the Docker Engine (and possibly other container
engines) won't accept to pull images from a registry that doesn't
use correct content types for manifest files, it could be argued
that it's better to give a hard fail. (Otherwise, the container
engine gives cryptic error messages like "missing signature key".)
I can change that behavior (and log errors but still store/serve
blobs to the filesystem) if you think it's better.
With https://github.com/google/nixery/pull/127, nixery will use extended
attributes to store metadata (when using local storage).
Right now, our integration test mounts a tmpfs to /var/cache/nixery.
However, *user* xattrs aren't supported with tmpfs [1], so setting
xattrs would fail.
To workaround this, use a folder in the current working directory and
hope it's backed by something supporting user xattrs (which is the case
for GitHub Actions).
[1]: https://man7.org/linux/man-pages/man5/tmpfs.5.html#NOTES
Drops the go2nix configuration in favour of pkgs.buildGoModule.
Note that the go.sum file is bloated by issues with cyclic
dependencies in some Google projects, but this large number of
dependencies is not actually built.
* remove a step that was not supposed to be committed ("Do we have
Docker?")
* remove setup of old temporary storage directory (now done in
integration script test instead)
* skip creation of out-link for initial Nixery build (to avoid
cache-busting on the second build)
In case the `GOOGLE_APPLICATION_CREDENTIALS` environment variable is not
set, a redirect to storage.googleapis.com is issued, which means the
underlying bucket objects need to be publicly accessible.
This wasn't really obvious until now, so further clarify it.
This copies the integration tests from `.travis.yaml` into a script,
documents the assumptions it makes, and wires it into GitHub Actions.
Contrary to the travis version, we don't use Nixery's GCS backend, as
handing out access to the bucket used, especially for PRs, needs to be
done carefully.
Adding back GCS to the integration test can be done at a later point,
either by using a mock server, or by only exposing the credentials for
master builds (and have the test script decide on whether
GOOGLE_APPLICATION_CREDENTIALS is set or not).
The previous travis version had some complicated post-mortem log
gathering - instead of doing this, we can just `docker run` nixery, but
fork it into the background with the shell - causing it to still be able
to log its output as it's running.
An additional `--rm` is appended, so the container gets cleaned up on
termination - this allows subsequent runs on non-CI infrastructure (like
developer laptops), without having to manually clean up containers.
Fixes#119.
We don't intend to label, authenticate or whatever with the
GITHUB_TOKEN, so there's not really a reason to give any broader
permissions than the defaults.
When serving a manifest, it is important to set the content-type
correctly (otherwise pulling an image is likely to give a cryptic
error message, "Error response from daemon: missing signature key").
This makes sure that we set the content-type properly for both
manifests and layers.
Extends storage.Persist to accept a Content-Type argument, which in
the GCS backend is persisted with the object to ensure that the object
is served back with this content-type.
This is not yet implemented for the filesystem backend, where the
parameter is simply ignored.
This should help in the case of clients which expect the returned
objects to have content-types set when, for example, fetching layers
by digest.
To ensure that registry clients which attempt to pull manifests by
their content hash can interact with Nixery, this change implements
persisting image manifests in the CAS in the same way as image layers.
In combination with the previous refactorings this means that Nixery's
serving flow is now compatible with containerd.
I have verified this locally, but CI currently only runs against
Docker and not containerd, which is something I plan to address in a
subsequent PR.
This fixes#102
Modifies the layer serving endpoint to be a generic blob-serving
endpoint that can handle both manifest and layer object "types".
Note that this commit does not yet populate the CAS with any
manifests.
Permission changes in the Travis CI Nix builders have caused this to
start failing, as the build user now has insufficient permissions to
use caches.
There may be a way to change the permissions instead, but in the
meantime we will just cause things to rebuild.
Channel serving has moved to a new subdomain, and the redirect
semantics have changed. Instead of serving temporary redirects,
permanent redirects are now issued.
I've reported this upstream as a bug, but this workaround will fix it
in the meantime.
This moves the pin from just being in the Travis configuration to also
being set in a nixpkgs-pin.nix file, which makes it trivial to build
at the right commit when performing local builds.
Previously, this was failing as follows:
```
these derivations will be built:
/nix/store/7rbrf06phkiyz31dwpq88x920zjhnw0c-nixery-popcount.drv
building '/nix/store/7rbrf06phkiyz31dwpq88x920zjhnw0c-nixery-popcount.drv'...
building
warning: GOPATH set to GOROOT (/nix/store/4859cp1v7zqcqh43jkqsayl4wrz3g6hp-go-1.13.4/share/go) has no effect
failed to initialize build cache at /homeless-shelter/.cache/go-build: mkdir /homeless-shelter: permission denied
builder for '/nix/store/7rbrf06phkiyz31dwpq88x920zjhnw0c-nixery-popcount.drv' failed with exit code 1
error: build of '/nix/store/7rbrf06phkiyz31dwpq88x920zjhnw0c-nixery-popcount.drv' failed
```
This gets rid of the package called "server" and instead moves
everything into the project root, such that Go actually builds us a
binary called `nixery`.
This is the first step towards factoring out CLI-based functionality
for Nixery.
The previous logic failed because single meta-packages such as
"nixery.dev/shell" would not end up removing the meta-package itself
from the list of packages passed to Nix, causing a build failure.
This was a regression introduced in 827468a.
Nixery itself is built with the buildLayeredImage system, which takes
some time to create large numbers of layers.
This adjusts the default number of image layers from 96 to 20.
Additionally Nixery's image is often loaded with `docker load -i`,
which ignores layer cache hits anyways.
Additionaly the CI build is configured to use only 1, which speeds up
CI runs.
Imports the package set twice in the builder expression: Once
configured for the target system, once configured for the native
system.
This makes it possible to fetch the actual image contents for the
required architecture, but use local tools to assemble the symlink
layer and metadata.
Specifying this meta-package toggles support for ARM64 images, for
example:
# Pull a default x86_64 image
docker pull nixery.dev/hello
# Pull an ARM64 image
docker pull nixery.dev/arm64/hello
