The C++ Agent provides the a complete implementation of the HTTP server required by the MTConnect standard. The agent provides the protocol and data collection framework that will work as a standalone server. Once built, you only need to specify the XML description of the devices and the location of the adapter.
Pre-built binary releases for Windows are available from Releases for those who do not want to build the agent themselves. For *NIX users, you will need libxml2, cppunit, and cmake as well as build essentials.
Version 2.1.0 Added MQTT Sink, Agent Restart and new JSON format (version 2)
Version 2.0.0 Rearchitecture of the agent with TLS, MQTT Adapter, Ruby Interpreter, Agent Adaptr, and much more
Version 1.7.0 added kinematics, solid models, and new specifications types.
Version 1.6.0 added coordinate systems, specifications, and tabular data.
Version 1.5.0 added Data Set capabilities and has been updated to use C++ 14.
Version 1.4.0 added time period filter constraint, compositions, initial values, and reset triggers.
Version 1.3.0 added the filter constraints, references, cutting tool archetypes, and formatting styles.
Version 1.2.0 added the capability to support assets.
Version 1.1.0.8 add the ability to run the C++ Agent as a Windows
service and support for a configuration file instead of command line
arguments. The agent can accept input from a socket in a pipe (|
)
delimited stream according to the descriptions given in the adapter
guide.
The current win32 binary is built statically and requires no additional dlls to run. This allows for a single exe distributable.
agent [help|install|debug|run] [configuration_file]
help Prints this message
install Installs the service
remove Remove the service
debug Runs the agent on the command line with verbose logging -
sets logging_level to debug
run Runs the agent on the command line
config_file The configuration file to load
Default: agent.cfg in current directory
When the agent is started without any arguments it is assumed it will be running as a service and will begin the service initialization sequence. The full path to the configuration file is stored in the registry in the following location:
\\HKEY_LOCAL_MACHINE\SOFTWARE\MTConnect\MTConnect Agent\ConfigurationFile
-
agent/
- This contains the application source files and CMake file. -
assets/
- Some example Cutting Tool asset files. -
lib/
- Third party library source. Contains source for cppunit, dlib, and libxml++. -
samples/
- Sample XML configuration files mainly used for tests. -
simulator/
- Ruby scripts to execute an adapter simulator, both command-line and log-replay. -
test/
- Various unit tests. -
tools/
- Ruby scripts to dump the agent and the adapter in SHDR format. Includes a sequence test script. -
unix/
- Unix init.d script -
win32/
- Libraries required only for the win32 build and the win32 solution.
The windows binary releases come with a prebuilt exe that is statically linked with the Microsoft Runtime libraries. Aside from the standard system libraries, the agent only requires winsock libraries. The agent has been test with version of Windows 2000 and later.
bin/
- Win32 binary (no dependencies required).
Platform specific instructions are at the end of the README.
The configuration file is using the standard Boost C++ file format. The configuration file format is flexible and allows for both many adapters to be served from one agent and an adapter to feed multiple agents. The format is:
Key = Value
The key can only occur once within a section and the value can be any
sequence of characters followed by a <CR>
. There no significance to
the order of the keys, so the file can be specified in free form. We
will go over some configurations from a minimal configuration to more
complex multi-adapter configurations.
Using a Files Configuration section, individual files or directories can be inserted into the request file space from the agent. To do this, we use the Files top level configuration declaration as follows:
Files {
schemas {
Path = ../schemas
Location = /schemas/
}
styles {
Path = ../styles
Location = /styles/
}
}
Each set of files must be declared using a named file description, like schema or
styles and the local Path
and the Location
the files will be mapped to in the
HTTP server namespace. For example:
http://example.com:5000/schemas/MTConnectStreams_2.0.xsd will map to ../schemas/MTConnectStreams_2.0.xsd
All files will be mapped and the directory names do not need to be the same. These files can be either served directly or can be used to extend the schema or add XSLT stylesheets for formatting the XML in browsers.
To specify the new schema for the documents, use the following declaration:
StreamsNamespaces {
e {
Urn = urn:example.com:ExampleStreams:2.0
Location = /schemas/ExampleStreams_2.0.xsd
}
}
This will use the ExampleStreams_2.0.xsd schema in the document. The e
is the alias that will be
used to reference the extended schema. The Location
is the location of the xsd file relative in
the agent namespace. The Location
must be mapped in the Files
section.
An optional Path
can be added to the ...Namespaces
declaration instead of declaring the
Files
section. The Files
makes it easier to include multiple files from a directory and will
automatically include all the default MTConnect schema files for the correct version.
(See SchemaVersion
option below)
You can do this for any one of the other documents:
StreamsNamespaces
DevicesNamespaces
AssetsNamespaces
ErrorNamespaces
The same can be done with style sheets, but only the Location is required.
StreamsStyle {
Location = /styles/Streams.xsl
}
An optional Path
can also be used to reference the xsl file directly. This will not
include other files in the path like css or included xsl transforms. It is advised to
use the Files
declaration.
The following can also be declared:
DevicesStyle
StreamsStyle
AssetsStyle
ErrorStyle
Here’s an example configuration file. The #
character can be used to
comment sections of the file. Everything on the line after the #
is
ignored. We will start with an extremely simple configuration file.
# A very simple file...
Devices = VMC-3Axis.xml
This is a one line configuration that specifies the XML file to load
for the devices. Since all the values are defaulted, an empty
configuration file can be specified. This configuration file will load
the VMC-3Axis.xml
file and try to connect to an adapter located on the
localhost at port 7878. VMC-3Axis.xml
must only contain the definition
for one device; if more devices exist, an error will be raised and the
process will exit.
Most configuration files will specify at least one adapter. The adapters are contained within a block. The Boost configuration file format allows for nested configurations and block associations. There are a number of configurations that can be given for each adapter. Multiple adapters can be specified for one device as well.
Devices = VMC-3Axis.xml
Adapters
{
VMC-3Axis
{
Host = 192.168.10.22
Port = 7878 # *Default* value...
}
}
This example loads the devices file as before, but specifies the list
of adapters. The device is taken from the name VMC-3Axis
that starts
the nested block and connects to the adapter on 192.168.10.22
and port
7878
. Another way of specifying the equivalent configuration is:
Devices = VMC-3Axis.xml
Adapters
{
Adapter_1
{
Device = VMC-3Axis
Host = 192.168.10.22
Port = 7878 # *Default* value...
}
}
Line 7 specifies the Device name associated with this adapter explicitly. We will show how this is used in the next example.
Multiple adapters can supply data to the same device. This is done by creating multiple adapter entries and specifying the same Device for each.
Devices = VMC-3Axis.xml
Adapters
{
Adapter_1
{
Device = VMC-3Axis
Host = 192.168.10.22
Port = 7878 # *Default* value...
}
# Energy sensor
Adapter_2
{
Device = VMC-3Axis
Host = 192.168.10.2
Port = 7878 # *Default* value...
}
}
Both Adapter_1
and Adapter_2
will feed the VMC-3Axis
device with
different data items. The Adapter_1
name is arbitrary and could just
as well be named EnergySensor
if desired as illustrated below.
Devices = VMC-3Axis.xml
Adapters
{
Controller
{
Device = VMC-3Axis
Host = 192.168.10.22
Port = 7878 # *Default* value...
}
EnergySensor
{
Device = VMC-3Axis
Host = 192.168.10.2
Port = 7878 # *Default* value...
}
}
In this example we change the port to 80 which is the default http port. This also allows HTTP PUT from the local machine and 10.211.55.2.
Devices = MyDevices.xml
Port = 80
AllowPutFrom = localhost, 10.211.55.2
Adapters
{
...
For browsers you will no longer need to specify the port to connect to.
If multiple devices are specified in the XML file, there must be an adapter feeding each device.
Devices = MyDevices.xml
Adapters
{
VMC-3Axis
{
Host = 192.168.10.22
}
HMC-5Axis
{
Host = 192.168.10.24
}
}
This will associate the adapters for these two machines to the VMC
and HMC devices in MyDevices.xml
file. The ports are defaulted to
7878, so we are not required to specify them.
In this example we demonstrate how to change the service name of the agent. This allows a single machine to run multiple agents and/or customize the name of the service. Multiple configuration files can be created for each service, each with a different ServiceName. The configuration file must be referenced as follows:
C:> agent install myagent.cfg
If myagent.cfg contains the following statements:
Devices = MyDevices.xml
ServiceName = MTC Agent 1
Adapters
{
...
The service will now be displayed as "MTC Agent 1" as opposed to "MTConnect Agent" and it will automatically load the contents of myagent.cfg with it starts. You can now use the following command to start this from a command prompt:
C:> net start "MTC Agent 1"
To remove the service, do the following:
C:> agent remove myagent.cfg
Logging configuration is specified using the logger_config
block. You
can change the logging_level
to specify the verbosity of the logging
as well as the destination of the logging output.
logger_config
{
logging_level = debug
output = file debug.log
}
This will log everything from debug to fatal to the file debug.log. For only fatal errors you can specify the following:
logger_config
{
logging_level = fatal
}
The default file is agent.log in the same directory as the agent.exe
file resides. The default logging level is info
. To have the agent log
to the command window:
logger_config
{
logging_level = debug
output = cout
}
This will log debug level messages to the current console window. When the agent is run with debug, it sets the logging configuration to debug and outputs to the standard output as specified above.
The MTConnect C++ Agent supports extensions by allowing you to specify your own XSD schema files. These files must include the MTConnect schema and the top level node is required to be MTConnect. The "x" in this case is the namespace. You MUST NOT use the namespace "m" since it is reserved for MTConnect. See example 9 for an example of changing the MTConnect schema file location.
There are four namespaces in MTConnect: Devices, Streams, Assets, and Error. In this example we will replace the Streams and Devices namespace with our own namespace so we can have validatable XML documents.
StreamsNamespaces {
x {
Urn = urn:example.com:ExampleStreams:1.2
Location = /schemas/ExampleStreams_1.2.xsd
Path = ./ExampleStreams_1.2.xsd
}
}
DevicesNamespaces {
x {
Urn = urn:example.com:ExampleDevices:1.2
Location = /schemas/ExampleDevices_1.2.xsd
Path = ./ExampleDevices_1.2.xsd
}
}
For each schema file we have three options we need to specify. The Urn is the urn in the schema file that will be used in the header. The Location is the path specified in the URL when requesting the schema file from the HTTP client and the Path is the path on the local file system.
If you only want to change the schema location of the MTConnect schema files and serve them from your local agent and not from the default internet location, you can use the namespace "m" and give the new schema file location. This MUST be the MTConnect schema files and the urn will always be the MTConnect urn for the "m" namespace -- you cannot change it.
StreamsNamespaces {
m {
Location = /schemas/MTConnectStreams_2.0.xsd
Path = ./MTConnectStreams_2.0.xsd
}
}
DevicesNamespaces {
m {
Location = /schemas/MTConnectDevices_2.0.xsd
Path = ./MTConnectDevices_2.0.xsd
}
}
The MTConnect agent will now serve the standard MTConnect schema files from the local directory using the schema path /schemas/MTConnectDevices_2.0.xsd.
We can also serve files from the MTConnect Agent as well. In this example we can assume we don't have access to the public internet and we would still like to provide the MTConnect streams and devices files but have the MTConnect Agent serve them up locally.
DevicesNamespaces {
x {
Urn = urn:example.com:ExampleDevices:2.0
Location = /schemas/ExampleDevices_2.0.xsd
Path = ./ExampleDevices_2.0.xsd
}
Files {
stream {
Location = /schemas/MTConnectStreams_2.0.xsd
Path = ./MTConnectStreams_2.0.xsd
}
device {
Location = /schemas/MTConnectDevices_2.0.xsd
Path = ./MTConnectDevices_2.0.xsd
}
}
Or use the short form for all files:
Files {
schemas {
Location = /schemas/MTConnectStreams_2.0.xsd
Path = ./MTConnectStreams_2.0.xsd
}
}
If you have specified in your xs:include schemaLocation inside the ExampleDevices_2.0.xsd file the location "/schemas/MTConnectStreams_2.0.xsd", this will allow it to be served properly. This can also be done using the Devices namespace:
DevicesNamespaces {
m {
Location = /schemas/MTConnectDevices_2.0.xsd
Path = ./MTConnectDevices_2.0.xsd
}
}
The MTConnect agent will allow you to serve any other files you wish as well. You can specify a new static file you would like to deliver:
Files {
myfile {
Location = /files/xxx.txt
Path = ./files/xxx.txt
}
The agent will not serve all files from a directory and will not provide an index function as this is insecure and not the intended function of the agent.
If the "-o with_ruby=True" build is selected, then use to following configuration:
Ruby {
module = path/to/module.rb
}
The module specified at the path given will be loaded. There are examples in the test/Resources/ruby directory in github: Ruby Tests.
The current functionality is limited to the pipeline transformations from the adapters. Future changes will include adding sources and sinks.
The following is a complete example for fixing the Execution of a machine:
class AlertTransform < MTConnect::RubyTransform
def initialize(name, filter)
@cache = Hash.new
super(name, filter)
end
@@count = 0
def transform(obs)
@@count += 1
if @@count % 10000 == 0
puts "---------------------------"
puts "> #{ObjectSpace.count_objects}"
puts "---------------------------"
end
dataItemId = obs.properties[:dataItemId]
if dataItemId == 'servotemp1' or dataItemId == 'Xfrt' or dataItemId == 'Xload'
@cache[dataItemId] = obs.value
device = MTConnect.agent.default_device
di = device.data_item('xaxisstate')
if @cache['servotemp1'].to_f > 10.0 or @cache['Xfrt'].to_f > 10.0 or @cache['Xload'].to_f > 10
newobs = MTConnect::Observation.new(di, "ERROR")
else
newobs = MTConnect::Observation.new(di, "OK")
end
forward(newobs)
end
forward(obs)
end
end
MTConnect.agent.sources.each do |s|
pipe = s.pipeline
puts "Splicing the pipeline"
trans = AlertTransform.new('AlertTransform', :Sample)
puts trans
pipe.splice_before('DeliverObservation', trans)
end
-
BufferSize
- The 2^X number of slots available in the circular buffer for samples, events, and conditions.Default: 17 -> 2^17 = 131,072 slots.
-
CheckpointFrequency
- The frequency checkpoints are created in the stream. This is used for current with the at argument. This is an advanced configuration item and should not be changed unless you understand the internal workings of the agent.Default: 1000
-
Devices
- The XML file to load that specifies the devices and is supplied as the result of a probe request. If the key is not found the defaults are tried.Defaults: probe.xml or Devices.xml
-
DisableAgentDevice
- When the schema version is >= 1.7, disable the creation of the Agent device.Default: false
-
PidFile
- UNIX only. The full path of the file that contains the process id of the daemon. This is not supported in Windows.Default: agent.pid
-
ServiceName
- Changes the service name when installing or removing the service. This allows multiple agents to run as services on the same machine.Default: MTConnect Agent
-
Port
- The port number the agent binds to for requests.Default: 5000
-
ServerIp
- The server IP Address to bind to. Can be used to select the interface in IPV4 or IPV6.Default: 0.0.0.0
-
AllowPut
- Allow HTTP PUT or POST of data item values or assets.Default: false
-
AllowPutFrom
- Allow HTTP PUT or POST from a specific host or list of hosts. Lists are comma (,) separated and the host names will be validated by translating them into IP addresses.Default: none
-
JsonVersion
- JSON Printer format. Old format: 1, new format: 2Default: 2
-
LegacyTimeout
- The default length of time an adapter can be silent before it is disconnected. This is only for legacy adapters that do not support heartbeats.Default: 600
-
ReconnectInterval
- The amount of time between adapter reconnection attempts. This is useful for implementation of high performance adapters where availability needs to be tracked in near-real-time. Time is specified in milliseconds (ms).Default: 10000
-
IgnoreTimestamps
- Overwrite timestamps with the agent time. This will correct clock drift but will not give as accurate relative time since it will not take into consideration network latencies. This can be overridden on a per adapter basis.Default: false
-
PreserveUUID
- Do not overwrite the UUID with the UUID from the adapter, preserve the UUID in the Devices.xml file. This can be overridden on a per adapter basis.Default: true
-
SchemaVersion
- Change the schema version to a different version number.Default: 2.0
-
ConversionRequired
- Global default for data item units conversion in the agent. Assumes the adapter has already done unit conversion.Default: true
-
UpcaseDataItemValue
- Always converts the value of the data items to upper case.Default: true
-
MaxAssets
- The maximum number of assets the agent can hold in its buffer. The number is the actual count, not an exponent.Default: 1024
-
MonitorConfigFiles
- Monitor agent.cfg and Devices.xml files and restart agent if they change.Default: false
-
MinimumConfigReloadAge
- The minimum age of a config file before an agent reload is triggered (seconds).Default: 15
-
Pretty
- Pretty print the output with indententationDefault: false
-
ShdrVersion
- Specifies the SHDR protocol version used by the adapter. When greater than one (1), allows multiple complex observations, likeCondition
andMessage
on the same line. If it equials one (1), then any observation requiring more than a key/value pair need to be on separate lines. This is the default for all adapters.Default: 1
-
SuppressIPAddress
- Suppress the Adapter IP Address and port when creating the Agent Device ids and names. This applies to all adapters.Default: false
-
WorkerThreads
- The number of operating system threads dedicated to the AgentDefault: 1
The following parameters must be present to enable https requests. If there is no password on the certificate, TlsCertificatePassword
may be omitted.
-
TlsCertificateChain
- The name of the file containing the certificate chain created from signing authorityDefault: NULL
-
TlsPrivateKey
- The name of the file containing the private key for the certificateDefault: NULL
-
TlsDHKey
- The name of the file containing the Diffie–Hellman keyDefault: NULL
-
TlsCertificatePassword
- The password used when creating the certificate. If none was supplied, do not use.Default: NULL
-
TlsOnly
- Only allow secure connections, http requests will be rejectedDefault: false
-
TlsVerifyClientCertificate
- Request and verify the client certificate against root authoritiesDefault: false
-
TlsClientCAs
- ForTlsVerifyClientCertificate
, specifies a file that contains additional certificate authorities for verificationDefault: NULL
-
MqttHost
- IP Address or name of the MQTT BrokerDefault: 127.0.0.1
-
MqttPort
- Port number of MQTT BrokerDefault: 1883
-
MqttCaCert
- CA Certificate for MQTT TLS connection to the MTT BrokerDefault: NULL
-
MqttTls
- TLS Certificate for secure connection to the MQTT BrokerDefault: NULL
-
DeviceTopic
- Prefix for the Device Model topicDefault: MTConnect/Device/
-
ObservationTopic
- Prefix for the Streams events, samples, and conditionsDefault: MTConnect/Observation/
-
AssetTopic
- Prefix for the AssetsDefault: MTConnect/Asset/
-
Adapters
- Adapters begins a list of device blocks. If the Adapters are not specified and the Devices file only contains one device, a default device entry will be created with an adapter located on the localhost and port 7878 associated with the device in the devices file.Default: localhost 5000 associated with the default device
-
Device
- The name of the device that corresponds to the name of the device in the Devices file. Each adapter can map to one device. Specifying a "*" will map to the default device.Default: The name of the block for this adapter or if that is not found the default device if only one device is specified in the devices file.
-
Host
- The host the adapter is located on.Default: localhost
-
Port
- The port to connect to the adapter.Default: 7878
-
Manufacturer
- Replaces the manufacturer attribute in the device XML.Default: Current value in device XML.
-
Station
- Replaces the Station attribute in the device XML.Default: Current value in device XML.
-
SerialNumber
- Replaces the SerialNumber attribute in the device XML.Default: Current value in device XML.
-
UUID
- Replaces the UUID attribute in the device XML.Default: Current value in device XML.
-
AutoAvailable
- For devices that do not have the ability to provide available events, ifyes
, this sets theAvailability
to AVAILABLE upon connection.Default: no (new in 1.2, if AVAILABILITY is not provided for device it will be automatically added and this will default to yes)
-
AdditionalDevices
- Comma separated list of additional devices connected to this adapter. This provides availability support when one adapter feeds multiple devices.Default: nothing
-
FilterDuplicates
- If value isyes
, filters all duplicate values for data items. This is to support adapters that are not doing proper duplicate filtering.Default: no
-
LegacyTimeout
- length of time an adapter can be silent before it is disconnected. This is only for legacy adapters that do not support heartbeats. If heartbeats are present, this will be ignored.Default: 600
-
ReconnectInterval
- The amount of time between adapter reconnection attempts. This is useful for implementation of high performance adapters where availability needs to be tracked in near-real-time. Time is specified in milliseconds (ms). Defaults to the top level ReconnectInterval.Default: 10000
-
IgnoreTimestamps
- Overwrite timestamps with the agent time. This will correct clock drift but will not give as accurate relative time since it will not take into consideration network latencies. This can be overridden on a per adapter basis.Default: Top Level Setting
-
PreserveUUID
- Do not overwrite the UUID with the UUID from the adapter, preserve the UUID in the Devices.xml file. This can be overridden on a per adapter basis.*Default*: false
-
RealTime
- Boost the thread priority of this adapter so that events are handled faster.Default: false
-
RelativeTime
- The timestamps will be given as relative offsets represented as a floating point number of milliseconds. The offset will be added to the arrival time of the first recorded event.Default: false
-
ConversionRequired
- Adapter setting for data item units conversion in the agent. Assumes the adapter has already done unit conversion. Defaults to global.Default: Top Level Setting
-
UpcaseDataItemValue
- Always converts the value of the data items to upper case.Default: Top Level Setting
-
ShdrVersion
- Specifies the SHDR protocol version used by the adapter. When greater than one (1), allows multiple complex observations, likeCondition
andMessage
on the same line. If it equials one (1), then any observation requiring more than a key/value pair need to be on separate lines. Applies to only this adapter.Default: 1
-
SuppressIPAddress
- Suppress the Adapter IP Address and port when creating the Agent Device ids and names. Default: false
-
-
Url
- The URL of the source agent.http:
orhttps:
are accepted for the protocol. -
SourceDevice
– The Device name or UUID for the source of the data -
Count
– the number of items request during a single sampleDefault: 1000
-
Polling Interval
– The interval used for streaming or polling in millisecondsDefault: 500ms
-
Reconnect Interval
– The interval between reconnection attampts in millisecondsDefault: 10000ms
-
Use Polling
– Force the adapter to use polling instead of streaming. Only set totrue
if x-multipart-replace blocked.Default: false
-
Heartbeat
– The heartbeat interval from the serverDefault: 10000ms
-
logger_config
- The logging configuration section.-
logging_level
- The logging level:trace
,debug
,info
,warn
,error
, orfatal
.Default:
info
Note: when running Agent with
agent debug
,logging_level
will be set todebug
. -
output
- The output file or stream. If using a file, specify as:"file <filename>"
. cout and cerr can be used to specify the standard output and standard error streams. Defaults to the same directory as the executable.Default: file
adapter.log
-
max_size
- The maximum log file size. Suffix can be K for kilobytes, M for megabytes, or G for gigabytes. No suffix will default to bytes (B). Case is ignored.Default: 10M
-
max_index
- The maximum number of log files to keep.Default: 9
-
schedule
- The scheduled time to start a new file. Can be DAILY, WEEKLY, or NEVER.Default: NEVER
-
The principle adapter data format is a simple plain text stream separated by the pipe character |
. Every line except for commands starts with an optional timestamp in UTC. If the timestamp is not supplied the agent will supply a timestamp of its own taken at the arrival time of the data to the agent. The remainder of the line is a key followed by data – depending on the type of data item is being written to.
A simple set of events and samples will look something like this:
2009-06-15T00:00:00.000000|power|ON|execution|ACTIVE|line|412|Xact|-1.1761875153|Yact|1766618937
A line is a sequence of fields separated by |
. The simplest requires one key/value pair. The agent will discard any lines where the data is malformed. The end must end with a LF (ASCII 10) or CR-LF (ASCII 15 followed by ASCII 10) (UNIX or Windows conventions respectively). The key will map to the data item using the following items: the id
attribute, the name
attribute, and the CDATA
of the Source
element. If the key does not match it will be rejected and the agent will log the first time it fails. Different data items categories and types require a different number of tokens. The following rules specify the requirements for the adapter tokens:
If the value itself contains a pipe character |
the pipe must be escaped using a leading backslash \
. In addition the entire value has to be wrapped in quotes:
2009-06-15T00:00:00.000000|description|"Text with \| (pipe) character."
Conditions require six (6) fields as follows:
<timestamp>|<data_item_name>|<level>|<native_code>|<native_severity>|<qualifier>|<message>
For a complete description of these fields, see the standard. An example line will look like this:
2014-09-29T23:59:33.460470Z|htemp|WARNING|HTEMP|1|HIGH|Oil Temperature High
The next special format is the Message. There is one additional field, native_code, which needs to be included:
2014-09-29T23:59:33.460470Z|message|CHG_INSRT|Change Inserts
Time series data also gets special treatment, the count and optional frequency are specified. In the following example we have 10 items at a frequency of 100hz:
2014-09-29T23:59:33.460470Z|current|10|100|1 2 3 4 5 6 7 8 9 10
The data item name can also be prefixed with the device name if this adapter is supplying data to multiple devices. The following is an example of a power meter for three devices named device1
, device2
, and device3
:
2014-09-29T23:59:33.460470Z|device1:current|12|device2:current|11|device3:current|10
All data items follow the formatting requirements in the MTConnect standard for the vocabulary and coordinates like PathPosition.
A new feature introduced in version 1.4 is the ability to announce a reset has been triggered. If we have a part count named pcount
that gets reset daily, the new protocol is as follows:
2014-09-29T23:59:33.460470Z|pcount|0:DAY
To specify the duration of the static, indicate it with an @
sign after the timestamp as follows:
2014-09-29T23:59:33.460470Z@100.0|pcount|0:DAY
A new feature in version 1.5 is the DATA_SET
representation which allows for key value pairs to be given. The protocol is similar to time series where each pair is space delimited. The agent automatically removes duplicate values from the stream and allows for addition, deletion and resetting of the values. The format is as follows:
2014-09-29T23:59:33.460470Z|vars|v1=10 v2=20 v3=30
This will create a set of three values. To remove a value
2014-09-29T23:59:33.460470Z|vars|v2 v3=
This will remove v2 and v3. If text after the equal =
is empty or the =
is not given, the value is deleted. To clear the set, specify a resetTriggered
value such as MANUAL
or DAY
by preceeding it with a colon :
at the beginning of the line.
2014-09-29T23:59:33.460470Z|vars|:MANUAL
This will remove all the values from the current set. The set can also be reset to a specific set of values:
2014-09-29T23:59:33.460470Z|vars|:MANUAL v5=1 v6=2
This will remove all the old values from the set and set the current set. Values will accumulate when addition pairs are given as in:
2014-09-29T23:59:33.460470Z|vars|v8=1 v9=2 v5=10
This will add values for v8 and v9 and update the value for v5 to 10. If the values are duplcated they will be removed from the stream unless a resetTriggered
value is given.
2014-09-29T23:59:33.460470Z|vars|v8=1 v9=2 v5=0
Will be detected as a duplicate with respect to the previous values and will be removed. If a partial update is given and the other values are duplicates, then will be stripped:
2014-09-29T23:59:33.460470Z|vars|v8=1 v9=3 v5=10
Will be effectively the same as specifying:
2014-09-29T23:59:33.460470Z|vars|v9=2
And the streams will only have the one value when a sample is request is made at that point in the stream.
When the discrete
flag is set to true
in the data item, all change tracking is ignored and each set is treated as if it is new.
One can quote values using the following methods: "<text>"
, '<text>'
, and {<text>}
. Spaces and other characters can be included in the text and the matching character can be escaped with a \
if it is required as follows: "hello \"there\""
will yield the value: hellow "there"
.
A TABLE
representation is similar to the DATA_SET
that has a key value pair as the value. Using the encoding mentioned above, the following representations are allowed for tables:
<timestamp>|wpo|G53.1={X=1.0 Y=2.0 Z=3.0 s='string with space'} G53.2={X=4.0 Y=5.0 Z=6.0} G53.3={X=7.0 Y=8.0 Z=9 U=10.0}
Using the quoting conventions explained in the previous section, the inner content can contain quoted text and exscaped values. The value is interpreted as a key/value pair in the same way as the DATA_SET
. A table can be thought of as a data set of data sets.
All the reset rules of data set apply to tables and the values are treated as a unit.
Assets are associated with a device but do not have a data item they are mapping to. They therefore get the special data item name @ASSET@
. Assets can be sent either on one line or multiple lines depending on the adapter requirements. The single line form is as follows:
2012-02-21T23:59:33.460470Z|@ASSET@|KSSP300R.1|CuttingTool|<CuttingTool>...
This form updates the asset id KSSP300R.1 for a cutting tool with the text at the end. For multiline assets, use the keyword --multiline--
with a following unique string as follows:
2012-02-21T23:59:33.460470Z|@ASSET@|KSSP300R.1|CuttingTool|--multiline--0FED07ACED
<CuttingTool>
...
</CuttingTool>
--multiline--0FED07ACED
The terminal text must appear on the first position after the last line of text. The adapter can also remove assets (1.3) by sending a @REMOVE_ASSET@ with an asset id:
2012-02-21T23:59:33.460470Z|@REMOVE_ASSET@|KSSP300R.1
Or all assets can be removed in one shot for a certain asset type:
2012-02-21T23:59:33.460470Z|@REMOVE_ALL_ASSETS@|CuttingTool
Partial updates to assets is also possible by using the @UPDATE_ASSET@ key, but this will only work for cutting tools. The asset id needs to be given and then one of the properties or measurements with the new value for that entity. For example to update the overall tool length and the overall diameter max, you would provide the following:
2012-02-21T23:59:33.460470Z|@UPDATE_ASSET@|KSSP300R.1|OverallToolLength|323.64|CuttingDiameterMax|76.211
There are a number of commands that can be sent as part of the adapter stream. These change some dynamic elements of the device information, the interpretation of the data, or the associated default device. Commands are given on a single line starting with an asterisk *
as the first character of the line and followed by a : . They are as follows:
-
Specify the Adapter Software Version the adapter supports:
* adapterVersion: <version>
-
Set the calibration in the device component of the associated device:
* calibration: XXX
-
Tell the agent that the data coming from this adapter requires conversion:
* conversionRequired: <yes|no>
-
Specify the default device for this adapter. The device can be specified as either the device name or UUID:
* device: <uuid|name>
-
Set the description in the device header of the associated device:
* description: XXX
-
Set the manufacturer in the device header of the associated device:
* manufacturer: XXX
-
Specify the MTConnect Version the adapter supports:
* mtconnectVersion: <version>
-
Set the nativeName in the device component of the associated device:
* nativeName: XXX
-
Tell the agent that the data coming from this adapter would like real-time priority:
* realTime: <yes|no>
-
Tell the agent that the data coming from this adapter is specified in relative time:
* relativeTime: <yes|no>
-
Set the serialNumber in the device header of the associated device:
* serialNumber: XXX
-
Specify the version of the SHDR protocol delivered by the adapter. See
ShdrVersion
above:* shdrVersion: <version>
-
Set the station in the device header of the associated device:
* station: XXX
Any other command will be logged as a warning.
The agent and the adapter have a heartbeat that makes sure each is responsive to properly handle disconnects in a timely manner. The Heartbeat frequency is set by the adapter and honored by the agent. When the agent connects to the adapter, it first sends a * PING
and then expects the response * PONG <timeout>
where <timeout>
is specified in milliseconds. So if the following communications are given:
Agent:
* PING
Adapter:
* PONG 10000
This indicates that the adapter is expecting a PING
every 10 seconds and if there is no PING
, in 2x the frequency, then the adapter should close the connection. At the same time, if the agent does not receive a PONG
within 2x frequency, then it will close the connection. If no PONG
response is received, the agent assumes the adapter is incapable of participating in heartbeat protocol and uses the legacy time specified above.
Just as with the SHDR protocol, these messages must end with an LF (ASCII 10) or CR-LF (ASCII 15 followed by ASCII 10).
There are two configuration settings mentioned above: AllowPut
and AllowPutFrom
. AllowPut
alone will allow any process to use HTTP
POST
or PUT
to send data to the agent and modify values. To restrict this to a limited number of
machines, you can list the IP Addresses that are allowed to POST
data to the agent.
An example would be:
AllowPut = yes
AllowPutFrom = 192.168.1.72, 192.168.1.73
This will allow the two machines to post data to the MTConnect agent. The data can be either data item values or assets. The primary use of this capability is uploading assets from a process or even the command line using utilities like curl. I'll be using curl for these examples.
For example, with curl you can use the -d option to send data to the server. The data will be in standard form data format, so all you need to do is to pass the <data_item_name>=<data_item_value>
to set the values, as follows:
curl -d 'avail=AVAILABLE&program_1=XXX' 'http://localhost:5000/ExampleDevice'
By specifying the device at the end of the URL, you tell the agent which device to use for the POST. This will set the availability tag to AVAILABLE and the program to XXX:
<Availability dataItemId="dtop_3" timestamp="2015-05-18T18:20:12.278236Z" name="avail" sequence="65">AVAILABLE</Availability>
...
<Program dataItemId="path_51" timestamp="2015-05-18T18:20:12.278236Z" name="program_1" sequence="66">XXX</Program>
The full raw data being passed over looks like this:
=> Send header, 161 bytes (0xa1)
0000: POST /ExampleDevice HTTP/1.1
001e: User-Agent: curl/7.37.1
0037: Host: localhost:5000
004d: Accept: */*
005a: Content-Length: 29
006e: Content-Type: application/x-www-form-urlencoded
009f:
=> Send data, 29 bytes (0x1d)
0000: avail=AVAILABLE&program_1=XXX
== Info: upload completely sent off: 29 out of 29 bytes
== Info: HTTP 1.0, assume close after body
<= Recv header, 17 bytes (0x11)
0000: HTTP/1.0 200 OK
<= Recv header, 20 bytes (0x14)
0000: Content-Length: 10
<= Recv header, 24 bytes (0x18)
0000: Content-Type: text/xml
<= Recv header, 2 bytes (0x2)
0000:
<= Recv data, 10 bytes (0xa)
0000: <success/>
>== Info: Closing connection 0
This is using the --trace - to dump the internal data. The response will be a simple <success/>
or <fail/>
.
Any data item can be set in this fashion. Similarly conditions are set using the following syntax:
curl -d 'system=fault|XXX|1|LOW|Feeling%20low' 'http://localhost:5000/ExampleDevice'
One thing to note, the data and values are URL encoded, so the space needs to be encoded as a %20 to appear correctly.
<Fault dataItemId="controller_46" timestamp="2015-05-18T18:24:48.407898Z" name="system" sequence="67" nativeCode="XXX" nativeSeverity="1" qualifier="LOW" type="SYSTEM">Feeling Low</Fault>
Assets are posted in a similar fashion. The data will be taken from a file containing the XML for the content. The syntax is very similar to the other requests:
curl -d @B732A08500HP.xml 'http://localhost:5000/asset/B732A08500HP.1?device=ExampleDevice&type=CuttingTool'
The @... uses the named file to pass the data and the URL must contain the asset id and the device name as well as the asset type. If the type is CuttingTool or CuttingToolArchetype, the data will be parsed and corrected if properties are out of order as with the adapter. If the device is not specified and there are more than one device in this adapter, it will cause an error to be returned.
Programmatically, send the data as the body of the POST or PUT request as follows. If we look at the raw data, you will see the data is sent over verbatim as follows:
=> Send header, 230 bytes (0xe6)
0000: POST /asset/B732A08500HP.1?device=ExampleDevice&type=CuttingTool
0040: HTTP/1.1
004b: User-Agent: curl/7.37.1
0064: Host: localhost:5000
007a: Accept: */*
0087: Content-Length: 2057
009d: Content-Type: application/x-www-form-urlencoded
00ce: Expect: 100-continue
00e4:
== Info: Done waiting for 100-continue
=> Send data, 2057 bytes (0x809)
0000: (file data sent here, see below...)
== Info: HTTP 1.0, assume close after body
<= Recv header, 17 bytes (0x11)
0000: HTTP/1.0 200 OK
<= Recv header, 20 bytes (0x14)
0000: Content-Length: 10
<= Recv header, 24 bytes (0x18)
0000: Content-Type: text/xml
<= Recv header, 2 bytes (0x2)
0000:
<= Recv data, 10 bytes (0xa)
0000: <success/>
<success/>== Info: Closing connection 0
The file that was included looks like this:
<CuttingTool serialNumber="1 " toolId="B732A08500HP" timestamp="2011-05-11T13:55:22" assetId="B732A08500HP.1" manufacturers="KMT">
<Description>
Step Drill KMT, B732A08500HP Grade KC7315
Adapter KMT CV50BHPVTT12M375
</Description>
<CuttingToolLifeCycle>
<CutterStatus><Status>NEW</Status></CutterStatus>
<ProcessSpindleSpeed nominal="5893">5893</ProcessSpindleSpeed>
<ProcessFeedRate nominal="2.5">2.5</ProcessFeedRate>
<ConnectionCodeMachineSide>CV50 Taper</ConnectionCodeMachineSide>
<Measurements>
<BodyDiameterMax code="BDX">31.8</BodyDiameterMax>
<BodyLengthMax code="LBX" nominal="120.825" maximum="126.325" minimum="115.325">120.825</BodyLengthMax>
<ProtrudingLength code="LPR" nominal="155.75" maximum="161.25" minimum="150.26">158.965</ProtrudingLength>
<FlangeDiameterMax code="DF" nominal="98.425">98.425</FlangeDiameterMax>
<OverallToolLength nominal="257.35" minimum="251.85" maximum="262.85" code="OAL">257.35</OverallToolLength>
</Measurements>
<CuttingItems count="2">
<CuttingItem indices="1" manufacturers="KMT" grade="KC7315">
<Measurements>
<CuttingDiameter code="DC1" nominal="8.5" maximum="8.521" minimum="8.506">8.513</CuttingDiameter>
<StepIncludedAngle code="STA1" nominal="90" maximum="91" minimum="89">89.8551</StepIncludedAngle>
<FunctionalLength code="LF1" nominal="154.286" minimum="148.786" maximum="159.786">157.259</FunctionalLength>
<StepDiameterLength code="SDL1" nominal="9">9</StepDiameterLength>
<PointAngle code="SIG" nominal="135" minimum="133" maximum="137">135.1540</PointAngle>
</Measurements>
</CuttingItem>
<CuttingItem indices="2" manufacturers="KMT" grade="KC7315">
<Measurements>
<CuttingDiameter code="DC2" nominal="12" maximum="12.011" minimum="12">11.999</CuttingDiameter>
<FunctionalLength code="LF2" nominal="122.493" maximum="127.993" minimum="116.993">125.500</FunctionalLength>
<StepDiameterLength code="SDL2" nominal="9">9</StepDiameterLength>
</Measurements>
</CuttingItem>
</CuttingItems>
</CuttingToolLifeCycle>
</CuttingTool>
An example in ruby is as follows:
> require 'net/http'
=> true
> h = Net::HTTP.new('localhost', 5000)
=> #<Net::HTTP localhost:5000 open=false>
> r = h.post('/asset/B732A08500HP.1?type=CuttingTool&device=ExampleDevice', File.read('B732A08500HP.xml'))
=> #<Net::HTTPOK 200 OK readbody=true>
> r.body
=> "<success/>"
The agent build is dependent on the following utilities:
- C++ Compiler compliant with C++ 17
- git is optional but suggested to download source and update when changes occur
- cmake for build generator and testing
- python 3 and pip to support conan for dependency and package management
- ruby and rake for mruby to support building the embedded scripting engine [not required if -o with_ruby=False]
The MTConnect Agent uses the conan package manager to install dependencies:
python 3 and Conan Package Manager Downloads
Download the Windows installer for your platform and run the installer.
You also need git and ruby if you want to embed mruby.
CMake is installed as part of Visual Studio. If you are using Visual Studio for the build, use the bundled version. Otherwise download from CMake CMake.
Install dependencies from the downloads above. Make sure python, ruby, and cmake are in your path.
pip install --upgrade pip
pip install conan
Clone the agent to another directory:
git clone https://github.com/mtconnect/cppagent.git
Make a build subdirectory of cppagent
cd cppagent
conan export conan\mqtt_cpp
conan export conan\mruby
Make sure to setup the environment:
"C:\Program Files (x86)\Microsoft Visual Studio\2019\Professional\VC\Auxiliary\Build\vcvars64.bat"
or
"C:\Program Files (x86)\Microsoft Visual Studio\2019\Community\VC\Auxiliary\Build\vcvars64.bat"
conan install . -if build -s build_type=Release --build=missing -pr conan/profiles/vs64
conan build . -bf build
conan install . -if build -s build_type=Release --build=missing -pr conan/profiles/vs32
conan build . -bf build
The windows XP 140 XP toolchain needs to be installed under individual component in the Visual Studio 2019 installer.
conan install . -s build_type=Release --build=missing -pr conan/profiles/vsxp
conan build . -bf build
cpack -G ZIP
The minimum memory (main + swap) when building with on CPU is 3GB. Less than that will likely cause the build to fail.
If the build runs out of resources, there are two options, you can add swap or set the following environment variable:
export CONAN_CPU_COUNT=1
to instruct conan to not parallelize the builds. Some of the modules that include boost beast require significant resources to build.
sudo apt-get install build-essential python3.9 python3-pip git cmake ruby rake
python3.9 -m pip install conan
echo 'export PATH=$HOME/.local/bin:$PATH' >> .bashrc
git clone https://github.com/mtconnect/cppagent.git
cd cppagent
conan export conan/mqtt_cpp
conan export conan/mruby
conan install . -if build --build=missing -pr conan/profiles/gcc
conan build . -bf build
Install brew and xcode command line tools
/bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh)"
xcode-select —install
brew install git
brew install cmake
pip3 install conan
git clone https://github.com/mtconnect/cppagent.git
cd cppagent
conan export conan/mqtt_cpp
conan export conan/mruby
conan install . -if build --build=missing
conan build . -bf build
mkdir build & cd build
cmake -G Xcode ..
apk add g++ python3 cmake git linux-headers make perl ruby
gem install rake
python3 -m ensurepip
python3 -m pip install --upgrade pip
export PATH=~/.local/bin:$PATH
pip3 install conan
git clone https://github.com/mtconnect/cppagent.git
cd cppagent
conan export conan/mqtt_cpp/
conan export conan/mruby/
conan install . -if build -pr conan/profiles/gcc --build=missing
conan build . -bf build
This section assumes you have installed openssl and can use the command line. The subject of the certificate is only for testing and should not be used in production. This section is provided to support testing and verification of the functionality. A certificate provided by a real certificate authority should be used in a production process.
NOTE: The certificates must be generated with OpenSSL version 1.1.1 or later. LibreSSL 2.8.3 is not compatible with more recent version of SSL on raspian (Debian).
openssl req -x509 -nodes -sha256 -days 3650 -newkey rsa:3072 -keyout rootca.key -out rootca.crt -subj "/C=US/ST=State/L=City/O=Your Company, Inc./OU=IT/CN=serverca.org"
openssl genrsa -out user.key 3072
openssl req -new -sha256 -key user.key -out user.csr -subj "/C=US/ST=State/L=City/O=Your Company, Inc./OU=IT/CN=user.org"
openssl x509 -req -in user.csr -CA rootca.crt -CAkey rootca.key -CAcreateserial -out user.crt -days 3650
openssl dhparam -out dh2048.pem 3072
openssl verify -CAfile rootca.crt rootca.crt
openssl verify -CAfile rootca.crt user.crt
openssl req -x509 -nodes -sha256 -days 3650 -newkey rsa:3072 -keyout clientca.key -out clientca.crt -subj "/C=US/ST=State/L=City/O=Your Company, Inc./OU=IT/CN=clientca.org"
openssl genrsa -out client.key 3072
openssl req -new -key client.key -out client.csr -subj "/C=US/ST=State/L=City/O=Your Company, Inc./OU=IT/CN=client.org"
For client.cnf
basicConstraints = CA:FALSE
nsCertType = client, email
nsComment = "OpenSSL Generated Client Certificate"
subjectKeyIdentifier = hash
authorityKeyIdentifier = keyid,issuer
keyUsage = critical, nonRepudiation, digitalSignature, keyEncipherment
extendedKeyUsage = clientAuth, emailProtection
openssl x509 -req -in client.csr -CA clientca.crt -CAkey clientca.key -out client.crt -CAcreateserial -days 3650 -sha256 -extfile client.cnf
openssl verify -CAfile clientca.crt clientca.crt
openssl verify -CAfile clientca.crt client.crt