Commit graph

147 commits

Author SHA1 Message Date
Vincent Ambo
3a5db4f9f1 refactor(server): Load GCS signing key from service account key
The JSON file generated for service account keys already contains the
required information for signing URLs in GCS, thus the environment
variables for toggling signing behaviour have been removed.

Signing is now enabled automatically in the presence of service
account credentials (i.e. `GOOGLE_APPLICATION_CREDENTIALS`).
2019-10-27 13:58:04 +01:00
Vincent Ambo
cca835ae37 fix(build): Only take the first matching hash for source hashing
Some Nix download mechanisms will add a second hash in the store path,
which had been added to the source hash output (breaking argument
interpolation).
2019-10-11 14:23:41 +01:00
Vincent Ambo
1853c74998 refactor(server): Only compress symlink forest layer once
Instead of compressing & decompressing again to get the underlying tar
hash, use a similar mechanism as for store path layers for the symlink
layer and only compress it once while uploading.
2019-10-11 12:37:00 +01:00
Vincent Ambo
e22ff5d176 fix(server): Use uncompressed tarball hashes in image config
Docker expects hashes of compressed tarballs in the manifest (as these
are used to fetch from the content-addressable layer store), but for
some reason it expects hashes in the configuration layer to be of
uncompressed tarballs.

To achieve this an additional SHA256 hash is calculcated while
creating the layer tarballs, but before passing them to the gzip
writer.

In the current constellation the symlink layer is first compressed and
then decompressed again to calculate its hash. This can be refactored
in a future change.
2019-10-11 12:37:00 +01:00
Vincent Ambo
0693e371d6 feat(server): Apply GZIP compression to all image layers
This fixes #62
2019-10-11 12:37:00 +01:00
Vincent Ambo
bf2718cebb chore(build): Use separate GCS bucket for CI runs
This has become an issue recently with changes such as GZIP
compression, where CI runs no longer work because they conflict with
the production bucket for the public instance.
2019-10-11 12:37:00 +01:00
Vincent Ambo
d7ffbbdea4 refactor(server): Use logrus convenience functions for logs
Makes use of the `.WithError` and `.WithField` convenience functions
in logrus to simplify log statement construction.

This has the added benefit of making it easier to correctly log
errors.
2019-10-06 23:05:23 +01:00
Vincent Ambo
c1020754a2 fix(build-image): Import 'match' from builtins 2019-10-06 23:05:23 +01:00
Vincent Ambo
6f148f789f refactor(server): Convert existing log entries to structured format
This rewrites all existing log statements into the structured logrus
format. For consistency, all errors are always logged separately from
the primary message in a field called `error`.

Only the "info", "error" and "warn" severities are used.
2019-10-06 23:05:23 +01:00
Vincent Ambo
f77c93b6ae feat(server): Add log level to severity mapping
The output format now writes a `severity` field that follows that
format that should be recognised by Stackdriver Logging.
2019-10-06 23:05:23 +01:00
Vincent Ambo
6912658c72 feat(server): Use hash of Nixery source as version
Uses a hash of Nixery's sources as the version displayed when Nixery
launches or logs an error. This makes it possible to distinguish
between errors logged from different versions.

The source hashes should be reproducible between different checkouts
of the same source tree.
2019-10-06 23:05:23 +01:00
Vincent Ambo
95abb1bcde feat(server): Initial Stackdriver-compatible log formatter
This formatter has basic support for the Stackdriver Error Reporting
format, but several things are still lacking:

* the service version (preferably git commit?) needs to be included in
  the server somehow
* log streams should be split between stdout/stderr as that is how
  AppEngine (and several other GCP services?) seemingly differentiate
  between info/error logs
2019-10-06 23:05:23 +01:00
Vincent Ambo
0642f7044d fix(server): Amend package path for Go tooling compatibility
With these changes it is possible to keep Nixery in $GOPATH and build
the server in there, while still having things work correctly via Nix.
2019-10-06 23:05:23 +01:00
Vincent Ambo
f4bf3518f6 refactor(server): Replace log calls with logrus
This introduces a structured logging library that can be used (next
step) to attach additional metadata to log entries.
2019-10-06 23:05:23 +01:00
Vincent Ambo
d9b329ef59 refactor(server): Always include 'cacert' & 'iana-etc'
These two packages almost always end up being required by programs,
but people don't necessarily consider them.

They will now always be added and their popularity is artificially
inflated to ensure they end up at the top of the layer list.
2019-10-03 22:50:02 +01:00
Vincent Ambo
9bb6d0ae25 fix(server): Ensure build cache objects are written to GCS
Cache writes might not be flushed without this call.
2019-10-03 22:50:02 +01:00
Vincent Ambo
48a5ecda97 feat(server): Order layers in image manifest based on merge rating
Image layers in manifests are now sorted in a stable (descending)
order based on their merge rating, meaning that layers more likely to
be shared between images come first.

The reason for this change is Docker's handling of image layers on
overlayfs2: Images are condensed into a single representation on disk
after downloading.

Due to this Docker will constantly redownload all layers that are
applied in a different order in different images (layer order matters
in imperatively created images), based on something it calls the
'ChainID'.

Sorting the layers this way raises the likelihood of a long chain of
matching layers at the beginning of an image.

This relates to #39.
2019-10-03 22:50:02 +01:00
Vincent Ambo
0d820423e9 chore(build-image): Remove nixery-build-layers
This functionality has been rolled into the server component and is no
longer required.
2019-10-03 22:29:50 +01:00
Vincent Ambo
6b06fe27be feat(server): Implement creation of layer tarballs in the server
This will create, upload and hash the layer tarballs in one disk read.
2019-10-03 22:29:50 +01:00
Vincent Ambo
1124b8c236 fix(server): Do not invoke layer build if no layers are missing
This previously invoked a Nix derivation that spent a few seconds on
making an empty object in JSON ...
2019-10-03 13:21:04 +01:00
Vincent Ambo
feba42e409 feat(server): Fetch popularity data on launch
The last missing puzzle piece for #50!
2019-10-03 13:21:04 +01:00
Vincent Ambo
43a642435b feat(server): Reimplement local manifest cache backed by files
Implements a local manifest cache that uses the temporary directory to
cache manifest builds.

This is necessary due to the size of manifests: Keeping them entirely
in-memory would quickly balloon the memory usage of Nixery, unless
some mechanism for cache eviction is implemented.
2019-10-03 13:21:04 +01:00
Vincent Ambo
313e5d08f1 refactor(builder): Streamline layer creation & reintroduce caching
The functions used for layer creation are now easier to follow and
have clear points at which the layer cache is checked and populated.

This relates to #50.
2019-10-03 13:21:04 +01:00
Vincent Ambo
53906024ff refactor: Remove remaining MD5-hash mentions and computations 2019-10-03 13:21:04 +01:00
Vincent Ambo
f6b40ed6c7 refactor(server): Cache manifest entries for layer builds
MD5 hash checking is no longer performed by Nixery (it does not seem
to be necessary), hence the layer cache now only keeps the SHA256 hash
and size in the form of the manifest entry.

This makes it possible to restructure the builder code to perform
cache-fetching and cache-populating for layers in the same place.
2019-10-03 13:21:04 +01:00
Vincent Ambo
355fe3f5ec feat(server): Reintroduce manifest caching to GCS
The new builder now caches and reads cached manifests to/from GCS. The
in-memory cache is disabled, as manifests are no longer written to
local file and the caching of file paths does not work (unless we
reintroduce reading/writing from temp files as part of the local
cache).
2019-10-03 13:21:04 +01:00
Vincent Ambo
1308a6e1fd refactor(server): Clean up cache implementation
A couple of minor fixes and improvements to the cache implementation.
2019-10-03 13:21:04 +01:00
Vincent Ambo
0698d7f2aa chore(server): Remove "layer seen" cache
This cache is no longer required as it is implicit because the layer
cache (mapping store path hashes to layer hashes) implies that a layer
has been seen.
2019-10-03 13:21:04 +01:00
Vincent Ambo
64fca61ea1 fix(server): Upload symlink layer created by first Nix build
This layer is needed in addition to those that are built in the second
Nix build.
2019-10-03 13:21:04 +01:00
Vincent Ambo
f4f2909573 fix(server): Specify correct authentication scope for GCS
When retrieving tokens for service service accounts, some methods of
retrieval require a scope to be specified.
2019-10-03 13:21:04 +01:00
Vincent Ambo
aa02ae1421 feat(server): Implement new build process core
Implements the new build process to the point where it can actually
construct and serve image manifests.

It is worth noting that this build process works even if the Nix
sandbox is enabled!

It is also worth nothing that none of the caching functionality that
the new build process enables (such as per-layer build caching) is
actually in use yet, hence running Nixery at this commit is prone to
doing more work than previously.

This relates to #50.
2019-10-03 13:21:04 +01:00
Vincent Ambo
17adda0355 fix(server): Minor fixes to updated new builder code 2019-10-03 13:21:04 +01:00
Vincent Ambo
ef2623d168 fix(nix): Minor fixes to derivations for new build process 2019-10-03 13:21:04 +01:00
Vincent Ambo
3f40c0a2d2 feat(server): Implement package for creating image manifests
The new manifest package creates image manifests and their
configuration. This previously happened in Nix, but is now part of the
server's workload.

This relates to #50.
2019-10-03 13:21:04 +01:00
Vincent Ambo
87e196757b feat(server): Reimplement creation & uploading of layers
The new build process can now call out to Nix to create layers and
upload them to the bucket if necessary.

The layer cache is populated, but not yet used.
2019-10-03 13:21:04 +01:00
Vincent Ambo
61269175c0 refactor(server): Introduce a state type to carry runtime state
The state type contains things such as the bucket handle and Nixery's
configuration which need to be passed around in the builder.

This is only added for convenience.
2019-10-03 13:21:04 +01:00
Vincent Ambo
6e2b84f475 feat(server): Add cache for layer builds in GCS & local cache
This cache is going to be used for looking up whether a layer build
has taken place already (based on a hash of the layer contents).

See the caching section in the updated documentation for details.

Relates to #50.
2019-10-03 13:21:04 +01:00
Vincent Ambo
6262dec8aa feat(nix): Add derivation to create layer tars from a store path set
This introduces a new Nix derivation that, given an attribute set of
layer hashes mapped to store paths, will create a layer tarball for
each of the store paths.

This is going to be used by the builder to create layers that are not
present in the cache.

Relates to #50.
2019-10-03 13:21:04 +01:00
Vincent Ambo
2c8ef634f6 docs(caching): Add information about Nixery's caching strategies 2019-10-03 13:21:04 +01:00
Vincent Ambo
e60805c9b2 feat(server): Introduce function to hash contents of a layer
This creates a cache key which can be used to check if a layer has
already been built.
2019-10-03 13:21:04 +01:00
Vincent Ambo
9c3c622403 refactor(server): Expose layer grouping logic via a function
Refactors the layer grouping package (which previously compiled to a
separate binary) to expose the layer grouping logic via a function
instead.

This is the next step towards creating layers inside of the server
component instead of in Nix.

Relates to #50.
2019-10-03 13:21:04 +01:00
Vincent Ambo
8c79d085ae chore(server): Import layer grouping logic into server component 2019-10-03 13:21:04 +01:00
Vincent Ambo
0898d8a961 chore(build-image): Simplify wrapper build & remove layer grouping
Simplifies the wrapper script used to invoke Nix builds from Nixery to
just contain the essentials, since the layer grouping logic is moving
into the server itself.
2019-10-03 13:21:04 +01:00
Vincent Ambo
712b38cbbc refactor(build-image): Do not assemble image layers in Nix
This is the first step towards a more granular build process where
some of the build responsibility moves into the server component.

Rather than assembling all layers inside of Nix, it will only create
the symlink forest and return information about the runtime paths
required by the image.

The server is then responsible for grouping these paths into layers,
and assembling the layers themselves.

Relates to #50.
2019-10-03 13:21:04 +01:00
Vincent Ambo
ad9b3eb262 refactor(build): Add group-layers to top-level Nix derivations
This makes CI build the group-layers tool (and cache it to Cachix!)
2019-09-22 18:04:40 +01:00
Vincent Ambo
7b987530d1 refactor(build-image): Minor tweak to layer construction script 2019-09-22 18:04:40 +01:00
Vincent Ambo
21a17b33f4 fix(build): Ensure launch script compatibility with other runtimes
Fixes two launch script compatibility issues with other container
runtimes (such as gvisor):

* don't fail if /tmp already exists
* don't fail if the environment becomes unset
2019-09-21 15:25:28 +01:00
Vincent Ambo
0000b956bb feat(server): Log Nix output live during the builds
Instead of dumping all Nix output as one at the end of the build
process, stream it live as the lines come in.

This is a lot more useful for debugging stuff like where manifest
retrievals get stuck.
2019-09-21 15:25:28 +01:00
Vincent Ambo
c391a7b7f8 fix(build-image): Use absolute paths in tarballs 2019-09-21 15:25:28 +01:00
Vincent Ambo
da6fd1d79e fix(build): Ensure nixery-build-image is on Nixery's PATH
This is useful when running Nixery locally.
2019-09-21 12:44:40 +01:00