Adjust the debouncing logic so that it applies to all inbound file
events, regardless of whether they match a sync or rebuild rule.
When the batch is flushed out, if any event for the service is a
rebuild event, then the service is rebuilt and all sync events for
the batch are ignored. If _all_ events in the batch are sync events,
then a sync is triggered, passing the entire batch at once. This
provides a substantial performance win for the new `tar`-based
implementation, as it can efficiently transfer the changes in bulk.
Additionally, this helps with jitter, e.g. it's not uncommon for
there to be double-writes in quick succession to a file, so even if
there's not many files being modified at once, it can still prevent
some unnecessary transfers.
Signed-off-by: Milas Bowman <milas.bowman@docker.com>
The IndexServerAddress field was as part of the initial Windows implementation
of the engine. For legal reasons, Microsoft Windows (and thus Docker images
based on Windows) were not allowed to be distributed through non-Microsoft
infrastructure. As a temporary solution, a dedicated "registry-win-tp3.docker.io"
registry was created to serve Windows images.
Using separate registries was not an ideal solution, and a more permanent
solution was created by introducing "foreign image layers" in the distribution
spec, after which the "registry-win-tp3.docker.io" ceased to exist, and
removed from the engine.
This replaces the code that calls out to the "/info" endpoint to use the
GetAuthConfigKey() function instead.
Related PR in docker/cli:
b4ca1c7368
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
When running `compose down`, the `--rmi` flag can be passed,
which currently supports two values:
* `local`: remove any _implicitly-named_ images that Compose
built
* `all` : remove any named images (locally-built or fetched
from a remote repo)
Removing images in the `local` case can be problematic, as it's
historically been done via a fair amount of inference over the
Compose model. Additionally, when using the "project-model"
(by passing `--project-name` instead of using a Compose file),
we're even more limited: if no containers for the project are
running, there's nothing to derive state from to perform the
inference on.
As a first pass, we started labeling _containers_ with the name
of the locally-built image associated with it (if any) in #9715.
Unfortunately, this still suffers from the aforementioned problems
around using actual state (i.e. the containers might no longer
exist) and meant that when operating in file mode (the default),
things did not behave as expected: the label is not available
in the project since it only exists at runtime.
Now, with these changes, Compose will label any images it builds
with project metadata. Upon cleanup during `down`, the engine
image API is queried for related images and matched up with the
services for the project. As a fallback for images built with
prior versions of Compose, the previous approach is still taken.
See also:
* https://github.com/docker/compose/issues/9655
* https://github.com/docker/compose/pull/9715
Signed-off-by: Milas Bowman <milas.bowman@docker.com>
The `compose-spec/compose-go` lib is written with `gopkg.in/yaml.v2`
as a target.
When marshalling via CLI (`compose convert` / `compose config`), we
were using a _different_ YAML lib, which was a fork of `go-yaml`,
which is what `gopkg.in/yaml.v2` is based off of.
Signed-off-by: Milas Bowman <milas.bowman@docker.com>
Milas Bowman
Guillaume Lours
Nicolas De Loof
Sebastiaan van Stijn
Lucas Berg
Ulysses Souza
Guillaume Lours
Mehrad Dadar