This moves to the newest etcd release, and also updates the imports to
the new go.etcd.io path. I think this is a bit of a pain, but might as
well get it done.
Send non-error log messages to stdout rather than stderr. Any messages
outside the main function are expected to be purely informational. By
sending to stdout rather than stderr, they can be discarded during the
build.
Fixes#568
I added a new feature in mkosi which got merged in:
31801e89e77188e697ed937ca6b8668fde4c4a4d
This allows us to pull in all of the git repository so that we can see
the version number that comes from git.
This commit adds a new consul:kv resource which allows us to set and
watch keys inside a consul kv datastore.
This was started by roidelapluie, and was finished during pair
programming with purpleidea.
Signed-off-by: Julien Pivotto <roidelapluie@inuits.eu>
Signed-off-by: James Shubin <james@shubin.ca>
This adds a "purge" parameter to the file resource. To do this, we have
to add the API hooks so the file resource can query other resources in
the graph to know if they are present, and as a result whether they
should be excluded from the purge or not.
This is useful for when we have a managed directory with some managed
contents. If a managed file is removed from the directory, then it will
be removed by the file (directory) resource if it has Purge set.
Alternatively, you can use the Reverse meta param, which is sometimes
preferable for this use case and sometimes not. This will be discussed
elsewhere.
This also adds a bunch of tests for this feature.
This also makes a few somewhat related cleanups in the file code.
This is a trait that specifies that your resource can be queried by
others. You can either enable it plainly to add wholesale access by
everyone, or you can implement your own Allow method to limit what is
permitted.
This adds a "fragments" mode to the file resource. In addition to
"content" and "source", you can now alternatively build a file from the
file fragments of other files.
Pretty cool that we have our first meme =D Have a look in the FAQ to see
the real reason we watch our resources. (TL;DR: It allows us to make
graph changes very fast.)
This unique token is necessary so that storing the files in the same dir
(basically a GitHub release) or in the SHA1SUMS file doesn't cause a
conflict.
This will be done by refactoring the current method, to return an error
message instead of a boolean value. This will also update a typo on the
user res.
