[ci] Manual trigger workflow to start Load testing

[appetrosyan]

Feature request

• Need a trigger with manual dispatch that takes the current iroha2-dev and iroha2-stable and deploys it on a k8s cluster with 4 peers. The peers are physically separated.

• Important Not all builds of the java SDK are compatible with all builds of iroha2-dev and sometimes not all builds of iroha2-stable.

• Then we deploy gatling as tuned by @timofeevmd, with the iroha-java SDK, and Grafana + Graphite.

Gatling should be pointed at the peers in the k8s cluster, the cluster must pass initial consensus and monitoring for all machines should be done to at least gather the

• RAM + CPU +

• optional prometheus metrics.

• Gatling also needs RAM + CPU monitoring.

• The Graphite needs to be ran on a separate machine to prevent a cascading failure.

Motivation

Need load testing to know about Iroha's performance metrics and behaviour under a stress load.

Who can help?

No response

[timofeevmd]

after talking with Vasily and Ekaterina, they came to the following implementation

postpone automatic deployment by commit

the product will be deployed manually - responsible Vasily Zyabkin

• Iroha should be deployed in k8s

Longevity booth remains in current configuration with current features

Deploy two virtual machines outside of k8s in the same geographic area as longevity stand

• 1 - load generator, Gatling load tool + JavaSDK deployed
  RAM 16
  CPU 4

• 2 - statistics, deployed Graphite
  RAM 6
  CPU 2

[BAStos525]

We have:

• five iroha2 pods in k8s cluster
• outside instance for longevity service run
• Ansible playbook + Jenkins job for longevity service deployment