Redis at Heroku Harold Giménez @hgmnz redisconf 2015 +

Redis at Heroku git deployment

proxy git push heroku master worker worker worker build service repos git push user

proxy git push heroku master worker worker worker build service repos build app breezing-swiftly-152 user

proxy git push heroku master worker worker worker build service repos I got this user

proxy git push heroku master worker worker worker build service repos user

proxy git push heroku master worker worker worker build service repos here’s my hostname and port user

proxy git push heroku master worker worker worker build service repos user

Redis at Heroku Logs

heroku logs --tail -a hgmnz 2010-09-16T15:13:46.677020+00:00 app[web.1]: Processing PostController#list (for 208.39.138.12 at 2010-09-16 15:13:46) [GET] 2010-09-16T15:13:46.677023+00:00 app[web.1]: Rendering template within layouts/application 2010-09-16T15:13:46.698234+00:00 app[web.1]: Completed in 74ms (View: 31, DB: 40) | 200 OK [http:// myapp.heroku.com/] 2010-09-16T15:13:46.723498+00:00 heroku[router]: at=info method=GET path="/posts" host=myapp.herokuapp.com" fwd="204.204.204.204" dyno=web.1 connect=1ms service=18ms status=200 bytes=975 2010-09-16T15:13:47.893472+00:00 app[worker.1]: 2 jobs processed at 16.6761 j/s, 0 failed ... $

app dyno 1 dyno 2 dyno 3 postgres primary postgres follower redis logplex config heroku logs --tail Logplex app log messages

Logplex canaries msg producer msg producer msg producer msg producer logging pipeline msg set message loss? latency?

Redis at Heroku Orchestration

Process orchestration Scaling web dynos... done, now running 10 Scaling worker dynos... done, now running 5 $ $ heroku ps:scale web=10 worker=5 -a demo-app process manager Messages sent to any Redis to boot,
 kill, restart processes dynos

Web dyno idling heroku router cluster keys with a 1h expiry HTTP traffic process manager which apps have had
 HTTP traffic
 in the last 24h?

Redis at Heroku Database Lifecycle

observe environment finite state machine

available creating uncertain unavailable deprovisioning deprovisioned

class Resource def feel observations.create( Feeler.new(self).current_environment ) end end class Feeler def current_environment { services_available?: service_available?, connections: connections, row_count: row_count, table_count: table_count, seq_scans: seq_scans, index_scans: index_scans } end end

class Resource include Stateful state :available do unless service_available? transition :uncertain end end end resource = Resource.find resource.transition :available resource.fetch_observation resource.tick puts resource.state # => :uncertain

module Stateful def self.included(base) base.extend ClassMethods end module ClassMethods def state(name, &block) state[name] = block end def states; @states ||={}; end end def tick self.instance_eval( &self.class.states[self.state.to_sym] ) end def transition(state); end end

resource.fetch_observation resource.tick Need to do this all the time

db1 db2 db3 db4 db5 db6 db7 db8 dbn W orkers db1.fetch_observation
 db1.tick

db db3 db2 db1 dbn Enqueue

Redis at Heroku Database Monitoring

Database Observations and Monitoring supervisor worker 1 worker 2 worker 3 worker 4 worker N I’m available for work

supervisor worker 1 worker 2 worker 3 worker 4 worker N Worker 1 monitor this DB Database Observations and Monitoring

supervisor worker 1 worker 2 worker 3 worker 4 worker N Database Observations and Monitoring monitor

supervisor worker 1 worker 2 worker 3 worker 4 worker N Database Observations and Monitoring

supervisor worker 1 worker 2 worker 3 worker 4 worker N I can’t reach this DB Database Observations and Monitoring

supervisor worker 1 worker 2 worker 3 worker 4 worker N Worker 4 monitor this DB Database Observations and Monitoring

supervisor worker 1 worker 2 worker 3 worker 4 worker N Database Observations and Monitoring monitor

worker Database Monitoring Last Observation cache postgres Heroku Postgres automation service GET /last_observation HSET db1 db_size 4213 HSET db1 cache_hit_rate 98

Redis at Heroku Other use cases

Router stats router node router node router node router node router node router node router node requests errors

pubsub + websockets dyno 1 dyno 2 dyno 3 pubsub browser websockets

I asked: what’s good and bad about Redis?

good bad •Fast •Simple •Powerful primitives •Expires on any data type •Many use cases, single infra •replication protocol is “surprisingly elegant” •No SSL support •Rotating creds is a pain •SPOF •Single threaded(ish)

Thanks! @hgmnz +