Home > docs > processes v2 > Configuration
The configuration
sections contains dependencies,
arguments and other process configuration values.
Process configuration
values can come from different sources: the section in
the concord.yml
file, request parameters, policies, etc. Here’s the order in
which all configuration
sources are merged before the process starts:
The runtime
parameter can be used to specify the execution runtime:
configuration:
runtime: "concord-v2"
Currently, the default runtime
is concord-v1
which is considered stable and
production-ready. This section describes the new and improved concord-v2
runtime that is available in “preview” mode - the syntax and the execution
semantics may change at any moment.
See the Processes (v1) section for more details
about concord-v1
runtime.
The entryPoint
configuration sets the name of the flow that will be used for
process executions. If no entryPoint
is specified the flow labelled default
is used automatically, if it exists.
configuration:
entryPoint: "main" # use "main" instead of "default"
flows:
main:
- log: "Hello World"
Note: some flow names have special meaning, such as onFailure
, onCancel
and onTimeout
. See the error handling section
for more details.
Default values for arguments can be defined in the arguments
section of the
configuration as simple key/value pairs as well as nested values:
configuration:
arguments:
name: "Example"
coordinates:
x: 10
y: 5
z: 0
flows:
default:
- log: "Project name: ${name}"
- log: "Coordinates (x,y,z): ${coordinates.x}, ${coordinates.y}, ${coordinates.z}"
Values of arguments
can contain expressions. Expressions can
use all regular tasks:
configuration:
arguments:
listOfStuff: ${myServiceTask.retrieveListOfStuff()}
myStaticVar: 123
Concord evaluates arguments in the order of definition. For example, it is possible to use a variable value in another variable if the former is defined earlier than the latter:
configuration:
arguments:
name: "Concord"
message: "Hello, ${name}"
A variable’s value can be defined or modified with the set step and a number of variables are automatically set in each process and available for usage.
The dependencies
array allows users to specify the URLs of dependencies such
as:
configuration:
dependencies:
# maven URLs...
- "mvn://org.codehaus.groovy:groovy-all:2.4.12"
# or direct URLs
- "https://repo1.maven.org/maven2/org/codehaus/groovy/groovy-all/2.4.12/groovy-all-2.4.12.jar"
- "https://repo1.maven.org/maven2/org/apache/commons/commons-lang3/3.6/commons-lang3-3.6.jar"
Concord downloads the artifacts and adds them to the process’ classpath.
Multiple versions of the same artifact are replaced with a single one, following standard Maven resolution rules.
Usage of the mvn:
URL pattern is preferred since it uses the centrally
configured list of repositories
and downloads not only the specified dependency itself, but also any required
transitive dependencies. This makes the Concord project independent of access
to a specific repository URL, and hence more portable.
Maven URLs provide additional options:
transitive=true|false
- include all transitive dependencies
(default true
);scope=compile|provided|system|runtime|test
- use the specific
dependency scope (default compile
).Additional options can be added as “query parameters” parameters to the dependency’s URL:
configuration:
dependencies:
- "mvn://com.walmartlabs.concord:concord-client:1.72.0?transitive=false"
The syntax for the Maven URL uses the groupId, artifactId, optionally packaging,
and version values - the GAV coordinates of a project. For example the Maven
pom.xml
for the Groovy scripting language runtime has the following
definition:
<project>
<groupId>org.codehaus.groovy</groupId>
<artifactId>groovy-all</artifactId>
<version>2.4.12</version>
...
</project>
This results in the path
org/codehaus/groovy/groovy-all/2.4.12/groovy-all-2.4.12.jar
in the
Central Repository and any repository manager proxying the repository.
The mvn
syntax uses the short form for GAV coordinates
groupId:artifactId:version
, so for example
org.codehaus.groovy:groovy-all:2.4.12
for Groovy.
Newer versions of groovy-all use <packaging>pom</packaging>
and define
dependencies. To use a project that applies this approach, called Bill of
Material (BOM), as a dependency you need to specify the packaging in between
the artifactId and version. For example, version 2.5.2 has to be specified as
org.codehaus.groovy:groovy-all:pom:2.5.2
:
configuration:
dependencies:
- "mvn://org.codehaus.groovy:groovy-all:pom:2.5.2"
The same logic and syntax usage applies to all other dependencies including Concord plugins.
A process can have a specific set of requirements
configured. Concord uses
requirements to control where the process should be executed and what kind of
resources it gets. For example, if the process specifies
configuration:
requirements:
agent:
favorite: true
and if there is an agent with
concord-agent {
capabilities = {
favorite = true
}
}
in its configuration file then it is a suitable agent for the process.
Following rules are used when matching requirements.agent
values of processes
and agent capabilities
:
capabilities
but missing in requirements.agent
is is ignored;capabilities
but present in requirements.agent
then it is not a match;requirements.agent
are treated as regular expressions,
i.e. in pseudo code capabilities_value.regex_match(requirements_value)
;requirements.agent
are treated as “one or more” match, i.e. if one
or more elements in the list must match the value from capabilities
;More examples:
configuration:
requirements:
agent:
size: ".*xl"
flavor:
- "vanilla"
- "chocolate"
matches agents with:
concord-agent {
capabilities = {
size = "xxl"
flavor = "vanilla"
}
}
You can specify the maximum amount of time the process can spend in
the RUNNING
state with the processTimeout
configuration. It can be useful
to set specific SLAs for deployment jobs or to use it as a global timeout:
configuration:
processTimeout: "PT1H"
flows:
default:
# a long running process
In the example above, if the process runs for more than 1 hour it is
automatically cancelled and marked as TIMED_OUT
.
The parameter accepts duration in the ISO 8601 format.
A special onTimeout
flow can be used to handle such processes:
flows:
onTimeout:
- log: "I'm going to run when my parent process times out"
The way Concord handles processTimeout
is described in more details in
the error handling
section.
Note: forms waiting for input and other processes in SUSPENDED
state
are not affected by the process timeout. I.e. a SUSPENDED
process can stay
SUSPENDED
indefinitely – up to the allowed data retention period.
The exclusive
section in the process configuration
can be used to configure
exclusive execution of the process:
configuration:
exclusive:
group: "myGroup"
mode: "cancel"
flows:
default:
- ${sleep.ms(60000)} # simulate a long-running task
In the example above, if another process in the same project with the same
group
value is submitted, it will be immediately cancelled.
If mode
set to wait
then only one process in the same group
is allowed to
run.
Note: this feature available only for processes running in a project.
The process event recording
can be configured using the events
section. Here is an example of the default
configuration:
configuration:
events:
recordTaskInVars: false
truncateInVars: true
recordTaskOutVars: false
truncateOutVars: true
truncateMaxStringLength: 1024
truncateMaxArrayLength: 32
truncateMaxDepth: 10
inVarsBlacklist:
- "apiKey"
- "apiToken"
- "password"
- "privateKey"
- "vaultPassword"
outVarsBlacklist: []
recordTaskInVars
, recordTaskOutVars
- enable or disable recording of
input/output variables in task calls;truncateInVars
, truncateOutVars
- if true
the runtime truncates
the recorded values to prevent spilling large values into process events;inVarsBlacklist
, outVarsBlacklist
- list of variable names that must
not be included recorded;truncateMaxStringLength
- maximum allowed length of string values.
The runtime truncates strings larger than the specified value;truncateMaxArrayLength
- maximum allowed length of array (list) values;truncateMaxDepth
- maximum allowed depth of nested data structures (e.g.
nested Map
objects).Note: in the runtime v1
the event recording configuration was a subsection of the runner
section.
In the runtime v2 it is a direct subsection of the configuration
block.