This can happen when you have --config-only network
Such attempt will fail anyway and it will create 15s delay in container
startup
Signed-off-by: Pavel Matěja <pavel@verotel.cz>
When a plugin has an activation error, it was not being checked in the
`waitActive` loop. This means it will just wait forever for a manifest
to be populated even though it may never come.
Signed-off-by: Brian Goff <cpuguy83@gmail.com>
c54b717 caused a regression for pluginv1 on Windows, where extraneous
backslashes were added to BasePath of the plugin. For pluginv1 on windows,
BasePath() should return an empty string, since the plugin is fully aware
of the mount path. Also, unlike Linux where all paths are relative to "/",
Windows paths are dependent on system drives and mapped drives.
Fixes#30148
Signed-off-by: Anusha Ragunathan <anusha.ragunathan@docker.com>
`plugins.GetAll()` was not locking the plugin map when checking if a
plugin exists, this can cause a race and potentially a panic if another
goroutine is trying to load a plugin into the map at the same time.
Also fixes a race during activation where a plugin inserts itself into
the plugin map but does not check if something else is already there.
This is already checked before trying to activate the plugin, however
the map lock is not held for this entire period, so other plugins may be
loaded during this time.
To fix, before inserting the plugin into the map, check if one with the
same name already exists and use that instead.
Signed-off-by: Brian Goff <cpuguy83@gmail.com>
When a plugin is activated, and then `plugins.Handle` is called to
register a new handler for a given plugin type, a deadlock occurs when
for anything which calls `waitActive`, including `Get`, and `GetAll`.
This happens because `Handle()` is setting `activated` to `false` to
ensure that plugin handlers are run on next activation.
Maybe these handlers should be called immediately for any plugins which
are already registered... but to preserve the existing behavior while
fixing the deadlock, track if handlers have been run on plugins and
reset when a new handler is registered.
The simplest way to reproduce the deadlock with Docker is to add a `-v
/foo` to the test container created for the external graphdriver tests.
Signed-off-by: Brian Goff <cpuguy83@gmail.com>
Legacy plugins expect host-relative paths (such as for Volume.Mount).
However, a containerized plugin cannot respond with a host-relative
path. Therefore, this commit modifies new volume plugins' paths in Mount
and List to prepend the container's rootfs path.
This introduces a new PropagatedMount field in the Plugin Config.
When it is set for volume plugins, RootfsPropagation is set to rshared
and the path specified by PropagatedMount is bind-mounted with rshared
prior to launching the container. This is so that the daemon code can
access the paths returned by the plugin from the host mount namespace.
Signed-off-by: Tibor Vass <tibor@docker.com>
Currently the plugins pkg allows a single handler. This assumption
breaks down if there are mutiple listeners to a plugin of a certain
Manifest such as NetworkDriver or IpamDriver when swarm-mode is enabled.
Signed-off-by: Madhu Venugopal <madhu@docker.com>
As part of making graphdrivers support pluginv2, a PluginGetter
interface was necessary for cleaner separation and avoiding import
cycles.
This commit creates a PluginGetter interface and makes pluginStore
implement it. Then the pluginStore object is created in the daemon
(rather than by the plugin manager) and passed to plugin init as
well as to the different subsystems (eg. graphdrivers, volumedrivers).
A side effect of this change was that some code was moved out of
experimental. This is good, since plugin support will be stable soon.
Signed-off-by: Anusha Ragunathan <anusha@docker.com>
Legacy plugin model maintained a map of plugins. This is
not used by the new model. Using this map in the new model
causes incorrect lookup of plugins. This change uses adds
a plugin to the map only if its legacy.
Signed-off-by: Anusha Ragunathan <anusha@docker.com>
This patch introduces a new experimental engine-level plugin management
with a new API and command line. Plugins can be distributed via a Docker
registry, and their lifecycle is managed by the engine.
This makes plugins a first-class construct.
For more background, have a look at issue #20363.
Documentation is in a separate commit. If you want to understand how the
new plugin system works, you can start by reading the documentation.
Note: backwards compatibility with existing plugins is maintained,
albeit they won't benefit from the advantages of the new system.
Signed-off-by: Tibor Vass <tibor@docker.com>
Signed-off-by: Anusha Ragunathan <anusha@docker.com>
When a plugin is first found, it is loaded into the available plugins
even though it's not activated yet.
If activation fails it is taken out of the list.
While it is in the list, other callers may see it and try to check it's
manifest. If it is not fully activated yet, the manifest will be nil and
cause a panic.
This is especially problematic for drivers that are down and have not
been activated yet.
We could just not load the plugin into the available list until it's
fully active, however that will just cause multiple of the same plugin
to attemp to be loaded.
We could check if the manifest is nil and return early (instead of
panicing on a nil manifest), but this will cause a 2nd caller to receive
a response while the first caller is still waiting, which can be
awkward.
This change uses a condition variable to handle activation (instead of
sync.Once). If the plugin is not activated, callers will all wait until
it is activated and receive a broadcast from the condition variable
signaling that it's ok to proceed, in which case we'll check if their
was an error in activation and proceed accordingly.
Signed-off-by: Brian Goff <cpuguy83@gmail.com>
Makes `docker volume ls` and `docker volume inspect` ask the volume
drivers rather than only using what is cached locally.
Previously in order to use a volume from an external driver, one would
either have to use `docker volume create` or have a container that is
already using that volume for it to be visible to the other volume
API's.
For keeping uniqueness of volume names in the daemon, names are bound to
a driver on a first come first serve basis. If two drivers have a volume
with the same name, the first one is chosen, and a warning is logged
about the second one.
Adds 2 new methods to the plugin API, `List` and `Get`.
If a plugin does not implement these endpoints, a user will not be able
to find the specified volumes as well requests go through the drivers.
Signed-off-by: Brian Goff <cpuguy83@gmail.com>
- Use the ones provided by docker/go-connections, they are a drop in replacement.
- Remove pkg/sockets from docker.
- Keep pkg/tlsconfig because libnetwork still needs it and there is a
circular dependency issue.
Signed-off-by: David Calavera <david.calavera@gmail.com>
This patches avoids registering (and calling) the same plugin more than
once. Using an helper map which indexes by name guarantees this and keeps
the order.
The behavior of overriding the same name in a flag is consistent with,
for instance, the `docker run -v /test -v /test` flag which register
the volume just once.
Adds integration tests.
Without this patch:
```
Dec 20 19:34:52 localhost.localdomain docker[9988]:
time="2015-12-20T19:34:52.080901676+01:00" level=debug msg="Calling
GET
/v1.22/info"
Dec 20 19:34:52 localhost.localdomain docker[9988]:
time="2015-12-20T19:34:52.081213202+01:00" level=debug msg="AuthZ
request using plugin docker-novolume-plugin"
Dec 20 19:34:52 localhost.localdomain docker[9988]:
time="2015-12-20T19:34:52.081268132+01:00" level=debug
msg="docker-novolume-plugin implements: authz"
Dec 20 19:34:52 localhost.localdomain docker[9988]:
time="2015-12-20T19:34:52.081699788+01:00" level=debug msg="AuthZ
request using plugin docker-novolume-plugin"
Dec 20 19:34:52 localhost.localdomain docker[9988]:
time="2015-12-20T19:34:52.081762507+01:00" level=debug
msg="docker-novolume-plugin implements: authz"
Dec 20 19:34:52 localhost.localdomain docker[9988]:
time="2015-12-20T19:34:52.082092480+01:00" level=debug msg="GET
/v1.22/info"
Dec 20 19:34:52 localhost.localdomain docker[9988]:
time="2015-12-20T19:34:52.628691038+01:00" level=debug msg="AuthZ
response using plugin docker-novolume-plugin"
Dec 20 19:34:52 localhost.localdomain docker[9988]:
time="2015-12-20T19:34:52.629880930+01:00" level=debug msg="AuthZ
response using plugin docker-novolume-plugin"
```
With this patch:
```
Dec 20 19:37:32 localhost.localdomain docker[16620]:
time="2015-12-20T19:37:32.376523958+01:00" level=debug msg="Calling
GET
/v1.22/info"
Dec 20 19:37:32 localhost.localdomain docker[16620]:
time="2015-12-20T19:37:32.376715483+01:00" level=debug msg="AuthZ
request using plugin docker-novolume-plugin"
Dec 20 19:37:32 localhost.localdomain docker[16620]:
time="2015-12-20T19:37:32.376771230+01:00" level=debug
msg="docker-novolume-plugin implements: authz"
Dec 20 19:37:32 localhost.localdomain docker[16620]:
time="2015-12-20T19:37:32.377698897+01:00" level=debug msg="GET
/v1.22/info"
Dec 20 19:37:32 localhost.localdomain docker[16620]:
time="2015-12-20T19:37:32.951016441+01:00" level=debug msg="AuthZ
response using plugin docker-novolume-plugin"
```
Also removes a somehow duplicate debug statement (leaving only the
second one as it's a loop of plugin's manifest):
```
Dec 20 19:52:30 localhost.localdomain docker[25767]:
time="2015-12-20T19:52:30.544090518+01:00" level=debug
msg="docker-novolume-plugin's manifest: &{[authz]}"
Dec 20 19:52:30 localhost.localdomain docker[25767]:
time="2015-12-20T19:52:30.544170677+01:00" level=debug
msg="docker-novolume-plugin implements: authz"
```
Signed-off-by: Antonio Murdaca <runcom@redhat.com>
This patch makes it such that plugin initialization is synchronized
based on the plugin name and not globally
Signed-off-by: Darren Shepherd <darren@rancher.com>
Check if there is a plugin socket first under `/run/docker/plugins/NAME.sock`.
If there is no socket for a plugin, check `/etc/docker/plugins/NAME.spec` and
`/usr/lib/docker/plugins/NAME.spec` for spec files.
Signed-off-by: David Calavera <david.calavera@gmail.com>
