Monitor a release, rollback to a previous version depending on the result of a PromQL (Prometheus), Lucene or DSL query (ElasticSearch).

$ helm plugin install https://github.com/ContainerSolutions/helm-monitor

A rollback happen only if the number of result from the query is greater than 0.

You can find a step-by-step example in the ./examples directory.

Monitor the peeking-bunny release against a Prometheus server, a rollback is initiated if the 5xx error rate is over 0 as measured over the last 5 minutes.

$ helm monitor prometheus peeking-bunny 'rate(http_requests_total{code=~"^5.*$"}[5m]) > 0'

You can connect to a given Prometheus instance, by default it will connect to http://localhost:9090.

$ helm monitor prometheus --prometheus=http://prometheus:9090 \
    peeking-bunny \
    'rate(http_requests_total{code=~"^5.*$"}[5m]) > 0'

Monitor the peeking-bunny release against an ElasticSearch server, a rollback is initiated if the 5xx error rate is over 0 for the last minute.

Using a Lucene query:

$ helm monitor elasticsearch peeking-bunny 'status:500 AND kubernetes.labels.app:app AND version:2.0.0'

Using a query DSL file:

$ helm monitor elasticsearch peeking-bunny ./query.json

You can connect to a given ElasticSearch instance, by default it will connect to http://localhost:9200.

$ helm monitor elasticsearch --elasticsearch=http://elasticsearch:9200 \
    peeking-bunny \
    'status:500 AND kubernetes.labels.app:app AND version:2.0.0'

Clone the repo, then add a symlink to the Helm plugin directory:

$ ln -s $GOPATH/src/github.com/ContainerSolutions/helm-monitor ~/.helm/plugins/helm-monitor

Install dependencies using dep:

$ dep ensure

Build:

$ go build -o helm-monitor ./cmd/...

Run:

$ helm monitor elasticsearch my-release ./examples/elasticsearch-query.json

• Investigate tool for failure detection compared to previous time window
• Add flag for fancy graph