This test shuts down two out of three managers and then asserts that the
swarm has a leader. A swarm that lost quorum won't necessarily have a
leader, and in this case only has one because the old leader is still
around. Soon SwarmKit will be changed so the leader gives up leadership
when quorum is lost. This will avoid confusing situations, like
read-only APIs succeeding, while ones that write to Raft hang.
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
Server-side rollback can take advantage of the rollback-specific update
parameters, instead of being treated as a normal update that happens to
go back to a previous version of the spec.
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
This adds support for placement preferences in Swarm services.
- Convert PlacementPreferences between GRPC API and HTTP API
- Add --placement-pref, --placement-pref-add and --placement-pref-rm to CLI
- Add support for placement preferences in service inspect --pretty
- Add integration test
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
This fix tries to address the issue raised in 30178 where
service healthcheck is `{}` in remote API will result in
dns resolve failue.
The reason was that when service healthcheck is `{}`,
service binding was not done.
This fix fixes the issue.
An integration test has been added.
This fix fixes 30178.
Signed-off-by: Yong Tang <yong.tang.github@outlook.com>
This fix tries to address the issue raised in 30242 where the `Scope`
field always changed to `swarm` in the ouput of `docker network ls/inspect`
when duplicate networks name exist.
The reason for the issue was that `buildNetworkResource()` use network name
(which may not be unique) to check for the scope.
This fix fixes the issue by always use network ID in `buildNetworkResource()`.
A test has been added. The test fails before the fix and passes after the fix.
This fix fixes 30242.
Signed-off-by: Yong Tang <yong.tang.github@outlook.com>
check to see if the node is part of a swarm, and if so, if it is unlocked first.
If neither of these are true, abort the command.
Signed-off-by: Ying Li <ying.li@docker.com>
… to make sure it doesn't fail. It also introduce StartWithError,
StopWithError and RestartWithError in case we care about the
error (and want the error to happen).
This removes the need to check for error and make the intent more
clear : I want a deamon with busybox loaded on it — if an error occur
it should fail the test, but it's not the test code that has the
responsability to check that.
Signed-off-by: Vincent Demeester <vincent@sbr.pm>
- use /secrets for swarm secret create route
- do not specify omitempty for secret and secret reference
- simplify lookup for secret ids
- do not use pointer for secret grpc conversion
Signed-off-by: Evan Hazlett <ejhazlett@gmail.com>
- fix lint issues
- use errors pkg for wrapping errors
- cleanup on error when setting up secrets mount
- fix erroneous import
- remove unneeded switch for secret reference mode
- return single mount for secrets instead of slice
Signed-off-by: Evan Hazlett <ejhazlett@gmail.com>
A recent change fixed integration tests to use "API" in test names
instead of "Api". A new test was added in a PR opened before this
change, and didn't benefit from the cleanup. Fix its name.
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
This adds support for two enhancements to swarm service rolling updates:
- Failure thresholds: In Docker 1.12, a service update could be set up
to either pause or continue after a single failure occurs. This adds
an --update-max-failure-ratio flag that controls how many tasks need to
fail to update for the update as a whole to be considered a failure. A
counterpart flag, --update-monitor, controls how long to monitor each
task for a failure after starting it during the update.
- Rollback flag: service update --rollback reverts the service to its
previous version. If a service update encounters task failures, or
fails to function properly for some other reason, the user can roll back
the update.
SwarmKit also has the ability to roll back updates automatically after
hitting the failure thresholds, but we've decided not to expose this in
the Docker API/CLI for now, favoring a workflow where the decision to
roll back is always made by an admin. Depending on user feedback, we may
add a "rollback" option to --update-failure-action in the future.
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
No substantial code change.
- Api --> API
- Cli --> CLI
- Http, Https --> HTTP, HTTPS
- Id --> ID
- Uid,Gid,Pid --> UID,PID,PID
- Ipam --> IPAM
- Tls --> TLS (TestDaemonNoTlsCliTlsVerifyWithEnv --> TestDaemonTLSVerifyIssue13964)
Didn't touch in this commit:
- Git: because it is officially "Git": https://git-scm.com/
- Tar: because it is officially "Tar": https://www.gnu.org/software/tar/
- Cpu, Nat, Mac, Ipc, Shm: for keeping a consistency with existing production code (not changable, for compatibility)
Signed-off-by: Akihiro Suda <suda.akihiro@lab.ntt.co.jp>
This fix tries to address the issue raised in 26090 where
remote API `POST /services/(id or name)/update` cannot
use `name` to update. This is not consistent with the
documentation of the remote API.
This fix fixes this issue by performing a lookup with `getService`
in case `name` instead of `id` is used in API.
This fix adds an integration test to cover the changes.
This fix fixes 26090.
Signed-off-by: Yong Tang <yong.tang.github@outlook.com>
Since 24237 has been merged, it is not necessary to require network
for swarm integration tests (`integration-cli/docker_api_swarm_test.go`)
any more.
This fix removes testRequires(c, Network) from swarm integration
tests.
This fix could be verified by disable networking, and all related
tests pass.
This fix is related to 24547, 24490, 24237.
This fix fixes 24547.
Signed-off-by: Yong Tang <yong.tang.github@outlook.com>
Ensure convergence before moving on with testing leader election
conditions. This reduce the flakiness of this test when run in different
environments.
Signed-off-by: Stephen J Day <stephen.day@docker.com>
If no restart delay is specified for a swarm service, the default
restart delay is 5 seconds. This is a reasonable value for actual
deployments - one example of where it's useful is that if a bad image is
specified, the orchestrator will wait 5 seconds between attempts to
restart it instead of restarting it in a tight loop.
In integration tests, this 5 second delay is dead time. The tests run
faster if the delay is reduced. Set it to 100 ms to avoid the waste of
time.
This appears to speed up a few tests:
DockerSwarmSuite.TestApiSwarmForceNewCluster 37.241s -> 34.323s
DockerSwarmSuite.TestApiSwarmRestartCluster 22.038s -> 15.545s
DockerSwarmSuite.TestApiSwarmServicesMultipleAgents 24.456s -> 19.853s
DockerSwarmSuite.TestApiSwarmServicesStateReporting 19.240s -> 10.049s
...a small step towards making the Swarm integration tests run in a
reasonable amount of time.
Also, change the update delay for the rolling update test from 8 seconds
to 4 seconds, which should be sufficient to differentiate between
batches of updated tasks. This reduces the runtime for
DockerSwarmSuite.TestApiSwarmServicesUpdate from 28s to 20s.
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
In `TestApiSwarmRestartCluster`, it's calling `checkClusterHealth`.
`checkClusterHealth` calls `d.info()`, which will return an error if
there is no cluster leader... problem is `checkClusterHealth` is doing a
nil error assertion w/o giving any time for a leader to be elected.
This moves the `d.info()` call into a `waitAndAssert` using the default
reconciliation timeout.
It also moves some other checks into a `waitAndAssert` to give the
cluster enough time to come back up.
Signed-off-by: Brian Goff <cpuguy83@gmail.com>
This changes the default behavior so that rolling updates will not
proceed once an updated task fails to start, or stops running during the
update. Users can use docker service inspect --pretty servicename to see
the update status, and if it pauses due to a failure, it will explain
that the update is paused, and show the task ID that caused it to pause.
It also shows the time since the update started.
A new --update-on-failure=(pause|continue) flag selects the
behavior. Pause means the update stops once a task fails, continue means
the old behavior of continuing the update anyway.
In the future this will be extended with additional behaviors like
automatic rollback, and flags controlling parameters like how many tasks
need to fail for the update to stop proceeding. This is a minimal
solution for 1.12.
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
There are currently problems with "swarm init" and "swarm join" when an
explicit --listen-addr flag is not provided. swarmkit defaults to
finding the IP address associated with the default route, and in cloud
setups this is often the wrong choice.
Introduce a notion of "advertised address", with the client flag
--advertise-addr, and the daemon flag --swarm-default-advertise-addr to
provide a default. The default listening address is now 0.0.0.0, but a
valid advertised address must be detected or specified.
If no explicit advertised address is specified, error out if there is
more than one usable candidate IP address on the system. This requires a
user to explicitly choose instead of letting swarmkit make the wrong
choice. For the purposes of this autodetection, we ignore certain
interfaces that are unlikely to be relevant (currently docker*).
The user is also required to choose a listen address on swarm init if
they specify an explicit advertise address that is a hostname or an IP
address that's not local to the system. This is a requirement for
overlay networking.
Also support specifying interface names to --listen-addr,
--advertise-addr, and the daemon flag --swarm-default-advertise-addr.
This will fail if the interface has multiple IP addresses (unless it has
a single IPv4 address and a single IPv6 address - then we resolve the
tie in favor of IPv4).
This change also exposes the node's externally-reachable address in
docker info, as requested by #24017.
Make corresponding API and CLI docs changes.
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
Implement the proposal from
https://github.com/docker/docker/issues/24430#issuecomment-233100121
Removes acceptance policy and secret in favor of an automatically
generated join token that combines the secret, CA hash, and
manager/worker role into a single opaque string.
Adds a docker swarm join-token subcommand to inspect and rotate the
tokens.
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
