in swarm.
Also, fix some CLI command confusions:
1. If the --external-ca flag is provided, require a --ca-cert flag as well, otherwise
the external CA is set but the CA certificate is actually rotated to an internal
cert
2. If a --ca-cert flag is provided, require a --ca-key or --external-ca flag be
provided as well, otherwise either the server will say that the request is
invalid, or if there was previously an external CA corresponding to the cert, it
will succeed. While that works, it's better to require the user to explicitly
set all the parameters of the new desired root CA.
This also changes the `swarm update` function to set the external CA's CACert field,
which while not strictly necessary, makes the CA list more explicit.
Signed-off-by: Ying Li <ying.li@docker.com>
parsing an environment file should give an error in case a zero-length
variable name (definition w/o a variable name) is encountered.
previously these lines went through unnoticed not informing the user about
a potential configuration error.
Signed-off-by: Tom Klingenberg <tklingenberg@lastflood.net>
previously docker did import environment variables if they were present
but created them if they were not when it was asked via a --env-file
cli option to import but not create them.
fix is to only import the variable into the environment if it is present.
additionally do not import variable names of zero-length (which are lines
w/ a potential variable definition w/o a variable name).
refs:
- https://github.com/docker/for-linux/issues/284
Signed-off-by: Tom Klingenberg <tklingenberg@lastflood.net>
test to show current behavior is wrong at parsing an environment file
defining an undefined variable - it must not be defined!
NOTE: this test assume the $HOME variable is always set (see POSIX, this
normally is the case, e.g. the test suite remains stable).
Signed-off-by: Tom Klingenberg <tklingenberg@lastflood.net>
'CheckInitialized' in the credential-helper library actually invokes
`pass`, which isn't desirable (see #699).
This moves the check to be simpler, and then pass will only be invoked
when it's needed (such as for `docker login` or when pulling from a
private registry).
This logic could also reasonably live in the credential-helper library,
but it's simple enough it seems fine in either location.
Signed-off-by: Euan Kemp <euank@euank.com>
Stores complete OCI descriptor instead of digest and platform
fields. This includes the size which was getting lost by not
storing the original manifest bytes.
Attempt to support existing cached files, if not output
the filename with the incorrect content.
Signed-off-by: Derek McGowan <derek@mcgstyle.net>
It makes it easier to get the correct stack from a compose config
struct without requiring the client (and thus talking to k8s API)
Signed-off-by: Vincent Demeester <vincent@sbr.pm>
- Handle `bundlefile` directly in the `top-level`
command. `bundlefile` is still experimental and will be deprecated
in future version — this should make be easier to remove it.
- Validate the `stack` name in all cases (i.e. whatever the
orchestrator is used)
- Load the composefile ahead of choosing the orchestrator. This
removes some slight duplication.
- Makes `RunDeploy` easier to use from outside packages (like
`docker/app`) with a preloaded configuration.
Signed-off-by: Vincent Demeester <vincent@sbr.pm>
- Add the possibility to skip interpolation
- Add the possibility to skip schema validation
- Allow customizing the substitution function, to add special cases.
Signed-off-by: Vincent Demeester <vincent@sbr.pm>
- `make test-e2e` runs the e2e tests twice : once against on
non-experimental daemon (as before), once against an experimental
daemon.
- adds `test-e2e-experimental` and `test-e2e-non-experimental` target
to run tests for the specified cases
Signed-off-by: Vincent Demeester <vincent@sbr.pm>