https://github.com/stealthly/punxsutawney

An Apache Mesos Framework that allows for replaying load over and over and over (and over) again
https://github.com/stealthly/punxsutawney

Last synced: 2 months ago
JSON representation

An Apache Mesos Framework that allows for replaying load over and over and over (and over) again

• Host: GitHub
• URL: https://github.com/stealthly/punxsutawney
• Owner: stealthly
• License: apache-2.0
• Created: 2015-05-25T18:53:24.000Z (about 9 years ago)
• Default Branch: master
• Last Pushed: 2015-08-10T16:34:02.000Z (almost 9 years ago)
• Last Synced: 2024-04-16T10:20:10.939Z (2 months ago)
• Language: Java
• Size: 215 KB
• Stars: 10
• Watchers: 12
• Forks: 2
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Lists

• awesome-mesos - Load replaying

README

        
# punxsutawney

An tool that allows for replaying load over and over and over (and over) again.

It uses Mesos/Marathon for running its HTTP clients.

It uses Kafka to read traffic data.

Primary components are:

- CLI - CLI to start HttpClients on Mesos (see Main);

- HttpClient - client running on Mesos, that generates HTTP traffic;

- Emitter - emitter (see SampleEmitter), producing traffic to Kafka topic (consumed by HttpClient(s));

# Running

In order to run the tool following steps should be executed:

1. Build via `./gradlew jar`

2. Start `./punxsutawney.sh`

3. Verify app is running via Mesos & Marathon UIs

4. Start emitting traffic to Kafka topic `punxsutawney` (see SampleEmitter)

5. Monitor stdout (via Mesos UI) of each HttpClient to see rps, failures, etc

Note: if running N HttpClient instances, Kafka topic should contain at least N partitions,

in order for HttpClients to be able to consume concurrently.

# Traffic generation

HttpClient consumes Kafka messages, containing serialized Requests instances.

Each Requests instance is designed to carry a batch of HTTP requests.

Emitter should batch 1k - 10k requests together to populate Requests object.