This standalone and runnable Spring Boot application communicates with automation subsystems, like a PLC or a Raspberry Pi. These kind of subsystems lack of resources and do most often not provide a higher level protocol on top of TCP/IP. Nowadays, PLC do also support protocols like MQTT, but this is used rarely.
In essence this module consists of Spring configuration to bootstrap the driver component in different configured flavors (see below). It is aware of multiple tenants (i.e. clients) and might run in the cloud with multiple instances and different configurations.
Note: Instantiating multiple instances of the driver component with same port settings on the same hosting machine is not possible, each driver instance blocks one TCP/IP socket (host/port) like configured in custom configuration. Each instance must have its own configuration, in particular its own TCP/IP port settings. A project (tenant) may have multiple drivers deployed, all running on different ports.
Find further documentation in the Wiki
A driver instance can be started in different operation modes: Simplex or Duplex communication with either Client or Server connection mode. All four can be combined in an arbitrary way.
In the simplex communication mode a client application connects to the driver with one socket for inbound and a seperate socket for the outbound communication. Sending and receiving messages is handled through separated and dedicated socket connections.
With the simplex communication mode the inbound and outbound communication modes are configured differently. By this the driver creates two
separate ConnectionFactories
, one for inbound and one for outbound. Each socket can be configured either in Client or Server mode.
A Client configured socket tries to connect to a listening server whereas the driver opens a listening socket when configured as Server.
A typical simplex configuration looks like:
owms:
driver:
connections:
hostname: 0.0.0.0
subsystems:
- name: SPS01
inbound:
mode: server
port: 30001
so-receive-buffer-size: 200
outbound:
mode: client
port: 30002
so-send-buffer-size: 200
identified-by-field: "RECV"
identified-by-value: "SPS01"
Each communication direction is configured with a different port setting and with mode set to server or client. To correlate outbound messages with previously received ones, the identified-by-* fields are used.
In contrast to simplex connection the driver instance can also be configured for bidirectional (duplex) mode where only one socket is used for inbound and outbound communication. Instead of configuring inbound or outbound only one duplex configuration is required.
owms:
driver:
connections:
subsystems:
- name: SPS03
duplex:
mode: server
hostname: localhost
port: 30003
so-send-buffer-size: 200
so-receive-buffer-size: 200
identified-by-field: "RECV"
identified-by-value: "SPS03"
The module uses a couple of well known Enterprise Integration Patterns (EIP), like Router, Transformer or the Enricher. For that reason Spring Integration is used as supporting integration framework. Additionally, this is a very convenient and flexible way to adopt new transport channels beside TCP/IP.
The overall integration architecture is shown below. The entry point is the inboundAdapter
that is connected to a TcpNetServerConnectionFactory
(not shown) and forwards incoming messages (synonym telegrams) to the inboundChannel
. A first transformer (telegramTransformer
)
terminates the ASCII string and converts it into a Spring Message.
This is done with support of the appropriate MessageMapper
that must exist for each telegram type. After the telegram is transformed into
a valid message type the generic messageRouter picks up the right queue and activates the proper ServiceActivator
.
Notice that the service activators queue name is built on the fly and follows a naming convention. This is one aspect to support requirement NR003.
The TCP/IP driver can run on locally, completely independent of cloud infrastructure services. The local setup is the appropriate deployment model during development or for small projects where no central infrastructure services are required. In a larger project setup with lots of subsystems and where the driver component is instantiated multiple times it makes sense to keep configuration on a central config server, this is where OpenWMS.org Configuration comes into play and requires to run in a distributed environment.
As already mentioned, all required configuration must be passed at startup time because no infrastructure services are required to access. Without any configuration the driver is started with the provided default configuration that is suitable for one driver instance:
owms:
driver:
server:
port: 30001
Property | Description |
---|---|
owms.driver.server.port | The unique port number the driver receives connections on. Multiple driver instances must have different portnumbers. |
In case you want to override the port number or other default configs at startup just set the environment variables accordingly. In the following example 2 drivers are instantiated, with different ports.
$ java -Dowms.driver.server.port=30001 -jar tcpip-driver-exec.jar
$ java -Dowms.driver.server.port=30002 -jar tcpip-driver-exec.jar
One could simply send an OSIP SYNQ telegram to the driver with port 30001 to get a response telegram:
$ telnet localhost 30001
Trying ::1...
Connected to localhost.
Escape character is '^]'.
###00160SPS01MFC__00001SYNQ20171123225959***********************************************************************************************************************
###00160MFC__SPS0100002SYNC20180927152848***********************************************************************************************************************
The first telegram string (SYNQ) is sent to the driver, whereas the driver responds with a SYNC telegram to synchronize the current system time.
In case of multiple driver components and lots of microservices it makes sense to keep service configuration at a central place and the OpenWMS.org Configuration server is the right choice. This infrastructure service takes the configuration of each process from a configured Git repository and passes it down to the processes at process startup. By using a configuration sever it is also possible to change configuration at runtime without the need to restart processes. Configuration is pushed down into the microservices and the service configuration is refreshed dynamically.
To run the driver component in a distributed fashion, the Spring profile "DISTRIBUTED" must be enabled. It's also a good practice to provide unique application names, at least when the driver is instantiated multiple times:
$ java -Dspring.profiles.active=DISTRIBUTED -Dspring.application.name=tcpip-palett1 -jar tcpip-driver-exec.jar
$ java -Dspring.profiles.active=DISTRIBUTED -Dspring.application.name=tcpip-palett2 -jar tcpip-driver-exec.jar
Each driver is now starting up and looking for a configuration server. The application name is used to load the appropriate driver configuration from. For example the Git repository KARL includes YAML configuration files for all processes used in KARL project, same is true for tcpip-palett1.yml and tcpip-palett2.yml.
tcpip-palett1.yml
owms:
tenant: KARL
driver:
server:
port: 30001
so-timeout: 300000
so-receive-buffer-size: 160
so-send-buffer-size: 160
routing-service-name: routing-service # is default
tcpip-palett2.yml
owms:
tenant: KARL
driver:
server:
port: 30002
so-timeout: 300000
so-receive-buffer-size: 160
so-send-buffer-size: 160
routing-service-name: routing-service # is default
The way how the driver communicates to other OpenWMS.org microservices can be defined by setting a Spring profile. In ASYNCHRONOUS mode the driver uses AMQP and sends the messages to RabbitMQ exchanges. In SYNCHRONOUS mode the driver calls defined REST endpoints of microservices.
Synchronous communication is used by default.
Either set the SYNCHRONOUS profile explicitly or omit it.
$ java -Dspring.profiles.active=SYNCHRONOUS -Dspring.application.name=tcpip-palett1 -jar tcpip-driver-exec.jar
To enable asynchronous communication over RabbitMQ set the Spring profile ASYNCHRONOUS
$ java -Dspring.profiles.active=ASYNCHRONOUS -Dspring.application.name=tcpip-palett1 -jar tcpip-driver-exec.jar
Important configuration properties of the driver component are the following.
Property | Description |
---|---|
owms.driver.timezone | The ZoneId (java.time.ZoneId) used to create timestamps |
owms.driver.serialization | (De-)Serialization method used for asynchronous communication. Possible values are barray and json |
owms.driver.osip.enabled | Whether OSIP telegram support is enabled or not |
owms.driver.osip.sync-field | Value of the SYNC field used to detect the start of a telegram |
owms.driver.osip.date-pattern | Date pattern used in OSIP telegrams |
owms.driver.routing-service.name | The logical service name of the TMS Routing Service |
owms.driver.routing-service.protocol | The protocol used to connect to the TMS Routing Service (eg. https) |
owms.driver.routing-service.username | The username for BASIC authentication |
owms.driver.routing-service.password | The password for BASIC authentication |
owms.driver.connections.hostname | The hostname setting inherited to all subsequent subsystem configurations |
owms.driver.connections.port-rest | The driver accepts incoming connections at this port in synchronous communication |
owms.driver.connections.so-timeout | The socket timeout inherited to all subsequent subsystem configurations |
owms.driver.connections.so-receive-buffer-size | The receiving buffer size inherited to all subsequent subsystem configurations |
owms.driver.connections.so-send-buffer-size | The sending buffer size inherited to all subsequent subsystem configurations |
owms.driver.connections.identified-by-field | The identified-by-field inherited to all subsequent subsystem configurations |
owms.driver.connections.subsystems | A list of subsystems. A driver can handle multiple subsystems in different modes |
owms.driver.connections.subsystems[].name | Unique name of the subsystem |
owms.driver.connections.subsystems[].inbound.mode | The operational mode. Either server or client |
owms.driver.connections.subsystems[].inbound.hostname | The hostname to connect to or the name of the interface to listen on (if mode is server ) |
owms.driver.connections.subsystems[].inbound.port | The port to connect to or to listen on |
owms.driver.connections.subsystems[].inbound.so-timeout | The socket timeout |
owms.driver.connections.subsystems[].inbound.so-receive-buffer-size | The size of the receiving buffer |
owms.driver.connections.subsystems[].outbound.mode | The operational mode. Either server or client |
owms.driver.connections.subsystems[].outbound.hostname | The hostname to connect to or the name of the interface to listen on (if mode is server ) |
owms.driver.connections.subsystems[].outbound.port | The port to connect to or to listen on |
owms.driver.connections.subsystems[].outbound.so-timeout | The socket timeout |
owms.driver.connections.subsystems[].outbound.so-send-buffer-size | The size of the send buffer |
owms.driver.connections.subsystems[].outbound.identified-by-field | The name of the telegram field that identifies the telegram receiver |
owms.driver.connections.subsystems[].outbound.identified-by-value | The actual telegram receiver name |
owms.driver.connections.subsystems[].duplex.mode | The operational mode. Either server or client |
owms.driver.connections.subsystems[].duplex.hostname | The hostname to connect to or the name of the interface to listen on (if mode is server ) |
owms.driver.connections.subsystems[].duplex.port | The port to connect to or to listen on |
owms.driver.connections.subsystems[].duplex.so-timeout | The socket timeout |
owms.driver.connections.subsystems[].duplex.so-send-buffer-size | The size of the send buffer |
owms.driver.connections.subsystems[].duplex.so-receive-buffer-size | The size of the receiving buffer |
owms.driver.connections.subsystems[].duplex.identified-by-field | The name of the telegram field that identifies the telegram receiver |
owms.driver.connections.subsystems[].duplex.identified-by-value | The actual telegram receiver name |
The driver component is configured for tenant aware logging. If, for instance, the tenant is configured to 'myProject' (owms.tenant=myProject) a tslog file (Technical Service Log) is written with name 'myProject-COMMON.tslog' that contains the time consumption processing each telegram took. An example tslog file looks like:
myProject COMMON 2018-09-27 15:26:27.280 INFO �[MEASURED ] : [TSL]>> ErrorMessageServiceActivator#wakeUp
myProject COMMON 2018-09-27 15:26:27.305 INFO �[MEASURED ] : [TSL]<< ErrorMessageServiceActivator#wakeUp took 26 [ms]
myProject COMMON 2018-09-27 15:27:01.586 INFO �[MEASURED ] : [TSL]>> TimesyncServiceActivator#wakeUp
myProject COMMON 2018-09-27 15:27:01.587 INFO �[MEASURED ] : [TSL]<< TimesyncServiceActivator#wakeUp took 1 [ms]
myProject COMMON 2018-09-27 15:28:48.638 INFO �[MEASURED ] : [TSL]>> TimesyncServiceActivator#wakeUp
myProject COMMON 2018-09-27 15:28:48.657 INFO �[MEASURED ] : [TSL]<< TimesyncServiceActivator#wakeUp took 22 [ms]
myProject COMMON 2018-09-27 15:28:50.704 INFO �[MEASURED ] : [TSL]>> TimesyncServiceActivator#wakeUp
Column | Description |
---|---|
#1 | Tenant name |
#2 | Module name (CORE, COMMON, TMS or WMS) |
#3 | Date of log entry written |
#4 | Time of log entry written |
#5 | Log level (INFO) |
#6 | Log category. All tslogs are using the category MEASURED |
#7 | TSL is another identifier used in log processing systems like logstash |
#8 | >>: incoming or <<: outgoing |
#9 | Type of telegram activator and indirectly the telegram handler used. By this information the processing telegram can be determined |
#10 | How long the message processing took in ms |
$ mvn deploy -Prelease,gpg -Ddebug.info=true