Requirements
This page describes StackState version 4.3.
The StackState 4.3 version range is End of Life (EOL) and no longer supported. We encourage customers still running the 4.3 version range to upgrade to a more recent release.
StackState can be installed on a Kubernetes cluster using the Helm charts provided by StackState. These charts have been tested and are compatible with Kubernetes 1.17.x, 1.18.x and 1.19.x (tested on Amazon EKS and Azure AKS), or the equivalent OpenShift release (version 4.4, 4.5 or 4.6) and Helm 3.
For a standard deployment, the StackState Helm chart will deploy storage services in a redundant setup with 3 instances of each service. The nodes required for different environments:
Recommended
Minimal
- Virtual machines: 5 nodes with
32GB memory
,8 vCPUs
- Amazon EKS: 5 instances of type
m5.2xlarge
orm4.2xlarge
- Azure AKS: 5 instances of type
D8s v3
orD8as V4
(Intel or AMD CPUs)
- Virtual machines: 4 nodes with
32GB memory
,8 vCPUs
- Amazon EKS: 4 instances of type
m5.2xlarge
orm4.2xlarge
- Azure AKS: 4 instances of type
D8s v3
orD8as V4
(Intel or AMD CPUs)
StackState uses persistent volume claims for the services that need to store data. The default storage class for the cluster will be used for all services unless this is overridden by values specified on the command line or in a
values.yaml
file. All services come with a pre-configured volume size that should be good to get you started, but can be customized later using variables as required.By default, the StackState Helm chart will deploy a router pod and service. This service's port
8080
is the only entry point that needs to be exposed via Ingress. You can access StackState without configuring Ingress by forwarding this port:kubectl port-forward service/<helm-release-name>-distributed-router 8080:8080
When configuring Ingress, make sure to allow for large request body sizes (50MB) that may be sent occasionally by data sources like the StackState Agent or the AWS integration.
It is not recommended to set a ResourceQuota as this can interfere with resource requests. The resources required by StackState will vary according to the features used, configured resource limits and dynamic usage patterns, such as Deployment or DaemonSet scaling.
If it is necessary to set a ResourceQuota for your implementation, the namespace resource limit should be set to match the node sizing requirements. For example, using the recommended node sizing for virtual machines (5 nodes with
32GB memory
, 8 vCPUs
), the namespace resource limit should be 5*32GB = 160GB
and 5*8 vCPUs = 40 vCPUs
.One of the following operating systems running Java. Check also the specific requirements for the StackState Agent StackPack:
OS | Release |
---|---|
Ubuntu | Bionic |
Ubuntu | Xenial |
Ubuntu | Trusty |
Fedora | 28 |
CentOS | 7 |
Debian | Stretch |
Red Hat | 7.5 |
Amazon Linux | 2 |
OpenJDK 8 patch level 121 or later.
StackState does not work with JDK versions 9 or higher at this time.
The StackState production setup runs on two machines and requires:
StackState node:
- 32GB of RAM
- 500GB disk space
- 8 cores CPU
StackGraph node:
- 24GB of RAM
- 500GB disk space
- 8 cores CPU
The POC setup runs on a single node and requires:
- 32GB of RAM
- 500GB disk space
- 8 cores CPU
The development setup runs on a single node and requires:
- 16GB of RAM
- 500GB disk space
- 4 cores CPU
To meet StackState minimal requirements, the AWS instance type needs to have at least:
- 4 CPU cores
- 16GB of memory, e.g., m5.xlarge.
The AWS CLI has to be installed on the EC2 instance that is running StackState.
Listed ports are TCP ports.
A production deployment separates StackState and StackState's database processes; StackGraph.
StackState has to be reachable on port 7070 by any supported browser. StackState port 7077 must be reachable from any system that is pushing data to StackState
StackGraph should be reachable by StackState on ports 2181, 8020, 15165, 16000, 16020, 50010.
The following ports can be opened for monitoring, but are also useful when troubleshooting:
- StackState: 9010, 9011, 9020, 9021, 9022, 9023, 9024, 9025, 9026
- StackGraph: 9001, 9002, 9003, 9004, 9005, 9006, 16010, 16030, 50070, 50075
StackState has to be reachable on port 7070 by any supported browser. StackState port 7077 must be reachable from any system that is pushing data to StackState
The following ports can be opened for monitoring, but are also useful when troubleshooting: 9001, 9002, 9003, 9004, 9005, 9006, 9010, 9011, 9020, 9021, 9022, 9023, 9024, 9025, 9026, 16010, 16030, 50070, 50075
Detailed information about ports per process.
PROCESS | PORT LIST |
---|---|
Elasticsearch | 9200: HTTP api 9300: Native api |
HBase Master | 16000: Master client API (needs to be open for clients) 16010: Master Web UI (optional) |
HBase Region Server | 16020: Region client API (needs to be open for clients) 16030: Region Web UI (optional) |
HDFS DataNode | 50010: Datanode API (needs to be open for clients) 50020: IPC api (communication within HDFS cluster) 50075: HTTP api (optional) |
HDFS NameNode | 8020: File system (needs to be open for clients) 50070: Web UI (optional) |
Kafka | 9092: Client port |
Receiver | 7077: HTTP agent API (aka receiver API). When using an agent, data is sent to this endpoint. |
StackGraph ProcessManager | 5152: StackGraph ProcessManager, at the moment only from localhost |
StackState | 7070: HTTP api & user interface 7071: Admin API for health checks and admin operations. Typically you want to use this only from `localhost` |
StackState ProcessManager | 5154: StackState ProcessManager, at the moment only from localhost |
Tephra Transaction service | 15165: Client API |
Zookeeper | 2181: Client API 2888: Zookeeper peers (general communication), only when running a cluster 3888: Zookeeper peers (leader election), only when running a cluster |
To use the StackState GUI, you must use one of the following web browsers:
- Chrome
- Firefox
Last modified 1yr ago