Docker swarm keeps a tasks.db file which can grow large...

Some of our deployments suddenly failed mysteriously: docker: Error response from daemon: OCI runtime create failed: container_linux.go:346: starting container process caused "process_linux.go:319: getting the final child's pid from pipe caused \"EOF\"": unknown.

Various commands showed something was wrong:

# docker service ls
insdns4o4bu8        xxx_1     replicated          0/1           xxx   *:8000->8000/tcp
dggj6i4actyn        xxx_2     replicated          0/1           xxx   *:8001->8001/tcp
# docker node ls
ID                            HOSTNAME            STATUS              AVAILABILITY        MANAGER STATUS
ux3enftr6krhlrp8bbodo2q1l *   wsdocker6           Down                Active              

systemctl status docker finally gave the right hint:
swarm component could not be started before timeout was reached

Which sent me to this issue comment which finally resolved the issue.

Turns out docker swarm has a tasks.db file which can get corrupted somehow.