TestSwarmLockUnlockCluster checked local node state immediately or
within one second after restarting managers. The daemon API can become
ready while the swarm node is still reconnecting, so the check could
fail as a Raft cluster converged.
Poll restart state with the existing reconciliation timeout and report
the last observed state on timeout. Reuse the wait for
TestSwarmJoinPromoteLocked, which crosses the same restart boundary.
The restarts remain because loading the encrypted node key from disk
is part of the behavior under test.
This addresses the node-state timeout only. Shutdown failures with
exit status 130 have a separate cause.
Signed-off-by: Paweł Gronowski <pawel.gronowski@docker.com>
Paweł Gronowski
06f0da3056