A simple docker container for automated stress testing of a graphite server using graphite-stresser for scheduled mode loading and haggar for accumulation mode loading
One mode must be chosen by setting either the scheduled mode, where a set of metrics is generated at set intervals and sent to the graphite server
# SCHEDULED MODE
GRAPHITE_STRESSER_MODE=scheduled
or the 'accumulate' mode, which gradually increases the number of agents sending metrics until the server is brought down.
# ACCUMULATION MODE
GRAPHITE_STRESSER_MODE=accumulate
There are 2 sets of environment variables, one set is applicable in both modes, the accumulation mode only ones are relevant only in that mode.
# BOTH SCHEDULED AND ACCUMULATION MODE
GRAPHITE_STRESSER_HOST: The graphite server host to target
GRAPHITE_STRESSER_PORT: The graphite server port to target
GRAPHITE_STRESSER_NUMHOSTS: The number of hosts to simulate sending data from
GRAPHITE_STRESSER_NUMTIMERS: The number of timers to create
# Possible timers values are 1, 2, 3, 4, 5, 10, 20, 64, 128, 256, 384, 650, 975, 1956, 3912, 4887, 7824, 9780, 13699]
GRAPHITE_STRESSER_INTERVAL: The metric publishing interval
GRAPHITE_STRESSER_DEBUG: True/false to enable or disable debug mode
GRAPHITE_STRESSER_LOGFILE: Optional logfile to send output to
# ACCUMULATION MODE
GRAPHITE_STRESSER_METRICS: The number of metrics for each agent to send
GRAPHITE_STRESSER_ACCUMULATION_INTERVAL: The interval to increase the agent count
GRAPHITE_STRESSER_JITTER: The jitter in metrics sent
GRAPHITE_STRESSER_AGENTS: The number of accumulation agents to run concurrently
GRAPHITE_STRESSER_HOST=localhost
GRAPHITE_STRESSER_PORT=2003
GRAPHITE_STRESSER_NUMHOSTS=1
GRAPHITE_STRESSER_NUMTIMERS=128
GRAPHITE_STRESSER_INTERVAL=10
GRAPHITE_STRESSER_DEBUG=false
GRAPHITE_STRESSER_METRICS=10000
GRAPHITE_STRESSER_ACCUMULATION_INTERVAL=10
GRAPHITE_STRESSER_JITTER=10
GRAPHITE_STRESSER_AGENTS=100
Metrics are published 15 at a time so for each publishing interval we will send
published_metrics_rate = timers * 15
published_metrics_per_minute = published_metrics_rate * (60/interval)
We then can plan a test set of messages and test the graphite server response
for example, below shows a run using the scheduled mode.
docker run --net=host -it --name graphite-stresser -e GRAPHITE_STRESSER_MODE=scheduled -e GRAPHITE_STRESSER_NUMTIMERS=128 registry.service.dsd.io/platforms/graphite-stresser:latest
and for accumulation mode
docker run --net=host -it --name graphite-stresser -e GRAPHITE_STRESSER_MODE=accumulate registry.service.dsd.io/platforms/graphite-stresser:latest
All the stress testers metrics are stored under the the keys Graphite.STRESS and Graphite.STRESS_ACCUMULATE and directories /graphite/whisper/STRESS and /graphite/whisper/STRESS_ACCUMULATE
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent