need to monitor at AppsFlyer? • Why we choose Sensu, which alternatives were considered? • Sensu architecture • Sensu configurations: server, clients, checks, handlers, mutators • Defining an alert, a simple demonstration • Our metrics flow, from collecting the metrics to getting alerts • Sensu API • Uchiwa dashboard • Mobile application • What Next?
for liveliness • System metrics • Services liveliness • Services applicative metrics • Performance • Dozens of 3rd party software, ex: Kafka, Rabbit, Couchbase, Tokumx and many more • That our service flow works correctly • That our exposed services are available from outside
Simple configuration • Natively passive • Standard check plugins • API • Interactive UI • Enable metric collection (Although we use collectd) • Secure message bus • Very fast implementation – ideal for growing startups
# in cluster Cluster # Env Region Domain "client": { "name": "<%= node[:fqdn] %>", "address": "<%= node[:fqdn] %>", "subscriptions": [ "base", "<%= node[:fqdn].to_s.split('-')[0] %>" ] } } ➢ Chef Template configuration use service name as subscriber { "checks": { "couchbase": { "handlers": ["pagerduty"], "command": "/opt/consul/scripts/chk_couchbase.py", "interval": 60, "occurrences": 2, "subscribers": [ "default","couchbase" ] } ➢ So, at the server when we configure alert we set “couchbase” as subscriber