DOI install resource availability
search cancel

DOI install resource availability

book

Article ID: 199284

calendar_today

Updated On:

Products

DX Operational Intelligence DX Application Performance Management

Issue/Introduction

 

DX OI 20.2 installation

Problem with installing the application layer on Kubernetes nodes.

How does the installer determine which nodes to use for the application layer? Does it take the ones that are not a master and not labeled for ElasticSearch (ES)?

In this case, there is one master node and three ES nodes. It was expected the application layer would be installed on the ES nodes as well.


========================= Hardware overview =============================
=========== Elasticsearch Layer - Required/Available/NodeName ============
[ PASSED ] Elasticsearch 1 - 1/23 CPU - node1
[ PASSED ] Elasticsearch 1 - 11/96 GB RAM - node1
[ PASSED ] Elasticsearch 2 - 1/23 CPU - node2
[ PASSED ] Elasticsearch 2 - 9/96 GB RAM - node2
[ PASSED ] Elasticsearch 3 - 1/23 CPU - node3
[ PASSED ] Elasticsearch 3 - 9/96 GB RAM - node3
============ Application Layer - Required/Available/NodeNames =====================
[ FAILED ] Application Layer Nodes - 25/0 CPU - []
[ FAILED ] Application Layer Nodes - 125/0 GB RAM - []
============ Unschedulable (not ready nodes) - CPU/RAM/NodeNames ==================
3 CPU/15 GB RAM - master
==========================================================================
[validate-resources] [ ERROR ] Hardware resources check failed. Add more resources to complete the installation.

Environment

CA DOI AO PLATFORM COMPONENTS 20.2.

Cause

 
 

Nodes labelled for use to host ElasticSearch are not included in installer resource calcuations, even if those nodes have enough resources to also host the application layer

In this situation, all nodes were labelled as ES nodes.

This is the label highlighted in bold on one node

node1    Ready    <none>   40h   v1.19.0   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,dxi-es-node=master-data-1,kubernetes.io/arch=amd64,kubernetes.io/hostname=node1,kubernetes.io/os=linux

 

The master could be configured to schedule pods but only for demo purposes where overall resources are limited.

Resolution

 

Allocate enough extra nodes to accommodate the application layer, or reduce labelled ES nodes to 1 if you are only deploying a test system

Additional Information

 

Information on ElasticSearch pre-requisites in Kubernetes including labelling

https://techdocs.broadcom.com/us/en/ca-enterprise-software/it-operations-management/dx-platform-on-premise/20-2/installing/set-up-the-environment/Kubernetes---Pre-installation-Tasks.html#concept.dita_8a89427a-18c1-47bb-beab-cc2cc0d3c934_ElasticsearchPrerequisites

Recommendation on separating application and ES layer

https://techdocs.broadcom.com/us/en/ca-enterprise-software/it-operations-management/dx-platform-on-premise/20-2/installing/software-requirements.html#concept.dita_d7ab03039be1dd1c0bbba340c2848b6d7d6349f2_HardwareRequirements