YARN allows applications to specify placement constraints in the form of data locality (preference to specific nodes or racks) or (non-overlapping) node labels. This document focuses on more expressive placement constraints in YARN. Such constraints can be crucial for the performance and resilience of applications, especially those that include long-running containers, such as services, machine-learning and streaming workloads.
For example, it may be beneficial to co-locate the allocations of a job on the same rack (affinity constraints) to reduce network costs, spread allocations across machines (anti-affinity constraints) to minimize resource interference, or allow up to a specific number of allocations in a node group (cardinality constraints) to strike a balance between the two. Placement decisions also affect resilience. For example, allocations placed within the same cluster upgrade domain would go offline simultaneously.
The applications can specify constraints without requiring knowledge of the underlying topology of the cluster (e.g., one does not need to specify the specific node or rack where their containers should be placed with constraints) or the other applications deployed. Currently, all constraints are hard, that is, if a constraint for a container cannot be satisfied due to the current cluster condition or conflicting constraints, the container request will remain pending or get rejected.
Note that in this document we use the notion of “allocation” to refer to a unit of resources (e.g., CPU and memory) that gets allocated in a node. In the current implementation of YARN, an allocation corresponds to a single container. However, in case an application uses an allocation to spawn more than one containers, an allocation could correspond to multiple containers.
We first describe how to enable scheduling with placement constraints and then provide examples of how to experiment with this feature using the distributed shell, an application that allows to run a given shell command on a set of containers.
To enable placement constraints, the following property has to be set to placement-processor or scheduler in conf/yarn-site.xml:
Property | Description | Default value |
---|---|---|
yarn.resourcemanager.placement-constraints.handler | Specify which handler will be used to process PlacementConstraints. Acceptable values are: placement-processor, scheduler, and disabled. | disabled |
We now give more details about each of the three placement constraint handlers:
The placement-processor handler supports a wider range of constraints and can allow more containers to be placed, especially when applications have demanding constraints or the cluster is highly-utilized (due to considering multiple containers at a time). However, if respecting task priority within an application is important for the user and the capacity scheduler is used, then the scheduler handler should be used instead.
Users can experiment with placement constraints by using the distributed shell application through the following command:
$ yarn org.apache.hadoop.yarn.applications.distributedshell.Client -jar share/hadoop/yarn/hadoop-yarn-applications-distributedshell-3.3.0.jar -shell_command sleep -shell_args 10 -placement_spec PlacementSpec
where PlacementSpec is of the form:
PlacementSpec => "" | PlacementExpr;PlacementSpec PlacementExpr => SourceTag,ConstraintExpr SourceTag => String(NumContainers) ConstraintExpr => SingleConstraint | CompositeConstraint SingleConstraint => "IN",Scope,TargetTag | "NOTIN",Scope,TargetTag | "CARDINALITY",Scope,TargetTag,MinCard,MaxCard | NodeAttributeConstraintExpr NodeAttributeConstraintExpr => NodeAttributeName=Value, NodeAttributeName!=Value CompositeConstraint => AND(ConstraintList) | OR(ConstraintList) ConstraintList => Constraint | Constraint:ConstraintList NumContainers => int Scope => "NODE" | "RACK" TargetTag => String MinCard => int MaxCard => int
Note:
An example of PlacementSpec is the following:
zk(3),NOTIN,NODE,zk:hbase(5),IN,RACK,zk:spark(7),CARDINALITY,NODE,hbase,1,3
The above encodes three constraints:
Another example below demonstrates a composite form of constraint:
zk(5),AND(IN,RACK,hbase:NOTIN,NODE,zk)
The above constraint uses the conjunction operator AND to combine two constraints. The AND constraint is satisfied when both its children constraints are satisfied. The specific PlacementSpec requests to place 5 “zk” containers in a rack where at least one “hbase” container is running, and on a node that no “zk” container is running. Similarly, an OR operator can be used to define a constraint that is satisfied when at least one of its children constraints is satisfied. Note that in case “zk” and “hbase” are containers belonging to different applications (which is most probably the case in real use cases), the allocation tags in the PlacementSpec should include namespaces, as we describe below (see Allocation tags namespace).
Allocation tags are string tags that an application can associate with (groups of) its containers. Tags are used to identify components of applications. For example, an HBase Master allocation can be tagged with “hbase-m”, and Region Servers with “hbase-rs”. Other examples are “latency-critical” to refer to the more general demands of the allocation, or “app_0041” to denote the job ID. Allocation tags play a key role in constraints, as they allow to refer to multiple allocations that share a common tag.
Note that instead of using the ResourceRequest object to define allocation tags, we use the new SchedulingRequest object. This has many similarities with the ResourceRequest, but better separates the sizing of the requested allocations (number and size of allocations, priority, execution type, etc.), and the constraints dictating how these allocations should be placed (resource name, relaxed locality). Applications can still use ResourceRequest objects, but in order to define allocation tags and constraints, they need to use the SchedulingRequest object. Within a single AllocateRequest, an application should use either the ResourceRequest or the SchedulingRequest objects, but not both of them.
Allocation tags might refer to containers of the same or different applications, and are used to express intra- or inter-application constraints, respectively. We use allocation tag namespaces in order to specify the scope of applications that an allocation tag can refer to. By coupling an allocation tag with a namespace, we can restrict whether the tag targets containers that belong to the same application, to a certain group of applications, or to any application in the cluster.
We currently support the following namespaces:
Namespace | Syntax | Description |
---|---|---|
SELF | self/${allocationTag} | The allocation tag refers to containers of the current application (to which the constraint will be applied). This is the default namespace. |
NOT_SELF | not-self/${allocationTag} | The allocation tag refers only to containers that do not belong to the current application. |
ALL | all/${allocationTag} | The allocation tag refers to containers of any application. |
APP_ID | app-id/${applicationID}/${allocationTag} | The allocation tag refers to containers of the application with the specified application ID. |
APP_TAG | app-tag/application_tag_name/${allocationTag} | The allocation tag refers to containers of applications that are tagged with the specified application tag. |
To attach an allocation tag namespace ns to a target tag targetTag, we use the syntax ns/allocationTag in the PlacementSpec. Note that the default namespace is SELF, which is used for intra-app constraints. The remaining namespace tags are used to specify inter-app constraints. When the namespace is not specified next to a tag, SELF is assumed.
The example constraints used above could be extended with namespaces as follows:
zk(3),NOTIN,NODE,not-self/zk:hbase(5),IN,RACK,all/zk:spark(7),CARDINALITY,NODE,app-id/appID_0023/hbase,1,3
The semantics of these constraints are the following:
The difference between allocation tags and node labels or node attributes, is that allocation tags are attached to allocations and not to nodes. When an allocation gets allocated to a node by the scheduler, the set of tags of that allocation are automatically added to the node for the duration of the allocation. Hence, a node inherits the tags of the allocations that are currently allocated to the node. Likewise, a rack inherits the tags of its nodes. Moreover, similar to node labels and unlike node attributes, allocation tags have no value attached to them. As we show below, our constraints can refer to allocation tags, as well as node labels and node attributes.
Applications can use the public API in the PlacementConstraints to construct placement constraint. Before describing the methods for building constraints, we describe the methods of the PlacementTargets class that are used to construct the target expressions that will then be used in constraints:
Method | Description |
---|---|
allocationTag(String... allocationTags) | Constructs a target expression on an allocation tag. It is satisfied if there are allocations with one of the given tags. |
allocationTagWithNamespace(String namespace, String... allocationTags) | Similar to allocationTag(String...), but allows to specify a namespace for the given allocation tags. |
nodePartition(String... nodePartitions) | Constructs a target expression on a node partition. It is satisfied for nodes that belong to one of the nodePartitions. |
nodeAttribute(String attributeKey, String... attributeValues) | Constructs a target expression on a node attribute. It is satisfied if the specified node attribute has one of the specified values. |
Note that the nodeAttribute method above is not yet functional, as it requires the ongoing node attributes feature.
The methods of the PlacementConstraints class for building constraints are the following:
Method | Description |
---|---|
targetIn(String scope, TargetExpression... targetExpressions) | Creates a constraint that requires allocations to be placed on nodes that satisfy all target expressions within the given scope (e.g., node or rack). For example, targetIn(RACK, allocationTag("hbase-m")), allows allocations on nodes that belong to a rack that has at least one allocation with tag “hbase-m”. |
targetNotIn(String scope, TargetExpression... targetExpressions) | Creates a constraint that requires allocations to be placed on nodes that belong to a scope (e.g., node or rack) that does not satisfy any of the target expressions. |
cardinality(String scope, int minCardinality, int maxCardinality, String... allocationTags) | Creates a constraint that restricts the number of allocations within a given scope (e.g., node or rack). For example, {@code cardinality(NODE, 3, 10, “zk”)} is satisfied on nodes where there are no less than 3 allocations with tag “zk” and no more than 10. |
minCardinality(String scope, int minCardinality, String... allocationTags) | Similar to cardinality(String, int, int, String...), but determines only the minimum cardinality (the maximum cardinality is unbound). |
maxCardinality(String scope, int maxCardinality, String... allocationTags) | Similar to cardinality(String, int, int, String...), but determines only the maximum cardinality (the minimum cardinality is 0). |
targetCardinality(String scope, int minCardinality, int maxCardinality, String... allocationTags) | This constraint generalizes the cardinality and target constraints. Consider a set of nodes N that belongs to the scope specified in the constraint. If the target expressions are satisfied at least minCardinality times and at most maxCardinality times in the node set N, then the constraint is satisfied. For example, targetCardinality(RACK, 2, 10, allocationTag("zk")), requires an allocation to be placed within a rack that has at least 2 and at most 10 other allocations with tag “zk”. |
The PlacementConstraints class also includes method for building compound constraints (AND/OR expressions with multiple constraints). Adding support for compound constraints is work in progress.
Applications have to specify the containers for which each constraint will be enabled. To this end, applications can provide a mapping from a set of allocation tags (source tags) to a placement constraint. For example, an entry of this mapping could be “hbase”->constraint1, which means that constraint1 will be applied when scheduling each allocation with tag “hbase”.
When using the placement-processor handler (see Enabling placement constraints), this constraint mapping is specified within the RegisterApplicationMasterRequest.
When using the scheduler handler, the constraints can also be added at each SchedulingRequest object. Each such constraint is valid for the tag of that scheduling request. In case constraints are specified both at the RegisterApplicationMasterRequest and the scheduling requests, the latter override the former.