Currently the daemon will not stop on error because the serve API job is
blocking the channel wait for daemon init. A better way is to run the
blocking serve API job as a goroutine and make sure that error
notification gets back to the main daemon thread (using the already
existing channel) so that clean shutdown can occur on error.
Docker-DCO-1.1-Signed-off-by: Phil Estes <estesp@linux.vnet.ibm.com> (github: estesp)
This fixes the `docker events`-related tests as they have been
failing due to clock skew between CI machine and test daemon on
some other machine (even 1-2 seconds of diff causes races as
we pass local time to --since/--until).
If we're running in same host, we keep using time.Now(), otherwise
we read the system time of the daemon from `/info` endpoint.
Fixes pretty much all events-related tests on windows CI.
Signed-off-by: Ahmet Alp Balkan <ahmetalpbalkan@gmail.com>
This change enables `fakeGIT()` to use the new `fakeStorage`
server which is automatically starting a container on the remote test
daemon machine using the git repo directory (when requested).
Fixes the following tests:
- `TestBuildApiLowerDockerfile`
- `TestBuildApiBuildGitWithF`
- `TestBuildApiDoubleDockerfile` (skipped on windows: NTFS case-insensitive)
- `TestBuildFromGIT` (still needs local server)
Signed-off-by: Ahmet Alp Balkan <ahmetalpbalkan@gmail.com>
Implemented a FakeStorage alternative that supports spinning
up a remote container on DOCKER_TEST_HOST to serve files over
an offline-compiled Go static web server image so that tests which
use URLs in Dockerfile can build them over at the daemon side.
`fakeStorage` function now automatically chooses if it should
use a local httptest.Server or a remote container.
This fixes the following tests when running against a remote
daemon:
- `TestBuildCacheADD`
- `TestBuildCopyWildcardNoFind`
- `TestBuildCopyWildcardCache`
- `TestBuildADDRemoteFileWithCache`
- `TestBuildADDRemoteFileWithoutCache`
- `TestBuildADDRemoteFileMTime`
- `TestBuildADDLocalAndRemoteFilesWithCache`
- `TestBuildADDLocalAndRemoteFilesWithoutCache`
- `TestBuildFromURLWithF`
- `TestBuildApiDockerFileRemote`
Signed-off-by: Ahmet Alp Balkan <ahmetalpbalkan@gmail.com>
Windows CI fails to dial remote test host over tcp in the test cases where
we clear environment variables during `exec(dockerBinary, ...)` in the
absence of `SystemRoot` environment variable (typically points to `c:\windows`).
This fixes tests:
- `TestRunEnvironmentErase`
- `TestRunEnvironmentOverride`
Signed-off-by: Ahmet Alp Balkan <ahmetalpbalkan@gmail.com>
Some pull/push tests are launching `registry-v2`
binary which is compiled and installed if the tests
are running in-container through `make test-integration-cli`.
However, registry is not supported to run on non-linux
platforms and we can't really spin up any registry-v2
containers in the remote DOCKER_TEST_HOST at this point.
Just skipping those with the new TestRequirement called
`RegistryHosting`.
Signed-off-by: Ahmet Alp Balkan <ahmetalpbalkan@gmail.com>
Shout out to @estesp for the idea. Some use cases of
`readContainerFile` can be replaced with `docker exec $id cat $file`.
This helper method can eliminate the requirement that
host/cli should be on the same machine.
TestRunMutableNetworkFiles and TestRunResolvconfUpdater still
need to access the docker host filesystem as they modify
the file directly from there assuming cli and daemon are
on the same machine.
This fixes TestLinksUpdateOnRestart and TestLinksHostsFilesInject
for Windows CI.
Signed-off-by: Ahmet Alp Balkan <ahmetalpbalkan@gmail.com>
Use `env -i` to very explicitly control exactly which environment variables leak into our tests. This enforces a clean separation of "build environment knobs" versus "test suite knobs".
This also includes a minor tweak to how we handle starting our integration daemon, especially to catch failure to start sooner than failing tests.
Signed-off-by: Andrew "Tianon" Page <admwiggin@gmail.com>
For Windows, we run integration-cli with DOCKER_TEST_HOST env var b/c
daemon is on some remote machine. This keeps the DOCKER_HOST set by
bash scripts in the env.
Signed-off-by: Ahmet Alp Balkan <ahmetalpbalkan@gmail.com>
TestGetContainersAttachWebsocket is currently broken on Windows CI tests
b/c it has hardcoded unix://var/run/docker.sock. This change makes use
of @icecrime's code in docker_utils and generalizes it with sockConn()
to provide a net.Conn by making use of DOCKER_TEST_HOST. Also fixes
the test.
Signed-off-by: Ahmet Alp Balkan <ahmetalpbalkan@gmail.com>
Applied multi parameters to pause and unpause.
Created a new test file dedicated for pause commands.
Created a new utility function to get a slice of paused containers.
Updated documentation
Signed-off-by: André Martins <martins@noironetworks.com>
I noticed that while we have tests to make sure that people don't
specify a Dockerfile (via -f) that's outside of the build context
when using the docker cli, we don't check on the server side to make
sure that API users have the same check done. This would be a security
risk.
While in there I had to add a new util func for the tests to allow us to
send content to the server that isn't json encoded - in this case a tarball
Signed-off-by: Doug Davis <dug@us.ibm.com>
Wait for the local registry-v2 test instance to become available to
avoid random tests failures.
Signed-off-by: Arnaud Porterie <arnaud.porterie@docker.com>
Tests no longer make the assumption that the daemon can be accessed
through unix:///var/run/docker.sock.
Signed-off-by: Arnaud Porterie <arnaud.porterie@docker.com>
Right now 'docker build' will send:
Sending build context to Docker daemon
to stderr, instead of stdout. This PR fixes that.
I looked in the rest of api/client/commands.go for other cases
that might do this and only one jumped out at me:
https://github.com/docker/docker/blob/master/api/client/commands.go#L2202
but I think if I changed that to go to stdout then it'll mess people up
who are expecting just the container ID to be printed to the screen and
there is no --quiet type of flag we can check.
Closes#9404
Signed-off-by: Doug Davis <dug@us.ibm.com>
Next steps, in another PR, would be:
- make all logging go through the logrus stuff
- I'd like to see if we can remove the env var stuff (like DEBUG) but we'll see
Closes#5198
Signed-off-by: Doug Davis <dug@us.ibm.com>
When a container is restarted all the volume configs are parsed again.
Even if the volume was already handled in a previous start it was still
calling "FindOrCreateVolume" on the volume repo causing a new volume to
be created.
This wasn't being detected because as part of the mount initialization
it checks to see if the the _mount_ was already initialized, but this
happens after the parsing of the configs.
So a check is added during parsing to skip a volume which was already
created for that container.
Signed-off-by: Brian Goff <cpuguy83@gmail.com>
