Table of Contents

Overview

...

• Data import jobs duration significantly increased compared to the Nolana results, see TestResults. It can be caused by added index on the ncp5 environment on Orchid release to make it production like. 
• Maximum CPU utilization was observed for mod-inventory-storage module during 80k DI job - 88%, see ServiceCPUUtilization.
• There is a significant memory utilization increase for some modules during 10k and 80k DI jobs, see MemoryUtilization. 
• Maximum service memory utilization reached 97% for mod-inventory-storage during 80k DI job, see MemoryUtilization.
• During 80k DI job there is a switch of instrances for reader and writer role. This might happen because of overload and restart of the writer instance, see DBCPUUtilization.

It should be noted that trigger process_marc_records_lb_insert_update_trigger was enabled for mod_source_record_storage.marc_records_lb table to make the environment production-like. It could influence the results in current release. At the same time tests showed that DI jobs duration with and without the trigger stays the same.

Recommendations & Jiras

Issue with failing 80K DI jobs was reproduced in 100% (tested several times). This issues requires further investigation.

Ticket: 

Test Results

Profile used for testing - Default - Create Holdings and SRS MARC Holdings

...

Infrastructure

PTF -environment ncp5 

• 11 m6i.2xlarge EC2 instances located in US East (N. Virginia)us-east-1  

• 2 database  instances, one reader, and one writer

  Name	API Name	Memory GIB	vCPUs	max_connections
  R6G Extra Large	db.r6g.xlarge	32 GiB	4 vCPUs	2731

  
  

• number of connections for mod-source-record-manager and mod-source-record-storage - 30 connections.
• MSK ptf-kakfa-3
  
  • 4 m5.2xlarge brokers in 2 zones

  • Apache Kafka version 2.8.0

  • EBS storage volume per broker 300 GiB

  • auto.create.topics.enable=true
  • log.retention.minutes=480
  • default.replication.factor=3
• Kafka topics partitioning: - 2 partitions for DI topics

...

512

...