Slide 1

Slide 1 text

Armin Ronacher iterate and ship

Slide 2

Slide 2 text

Armin Ronacher @mitsuhiko Flask / Sentry / Lektor

Slide 3

Slide 3 text

̿ SENTRY ̀

Slide 4

Slide 4 text

No content

Slide 5

Slide 5 text

No content

Slide 6

Slide 6 text

No content

Slide 7

Slide 7 text

̿ THE TWO PRODUCTS ̀

Slide 8

Slide 8 text

sentry vs ‘getsentry’

Slide 9

Slide 9 text

sentry open source repo on-premise monthly releases

Slide 10

Slide 10 text

‘getsentry’ billing & quotas depends on sentry hourly deploys

Slide 11

Slide 11 text

̿ THE GOALS ̀

Slide 12

Slide 12 text

deploy in seconds be unable to screw up and if you do: instant rollbacks

Slide 13

Slide 13 text

tag a release once a month

Slide 14

Slide 14 text

̿ WORKFLOW ̀

Slide 15

Slide 15 text

commit review integration deploy

Slide 16

Slide 16 text

requires good test coverage requires good local setup makes it easier for newcomers

Slide 17

Slide 17 text

̿ COMMITTING ̀

Slide 18

Slide 18 text

lint on commit!

Slide 19

Slide 19 text

1 Release / Month 5 Deployments / Day On Prem: Hosted:

Slide 20

Slide 20 text

No content

Slide 21

Slide 21 text

master is stable

Slide 22

Slide 22 text

1. branch off master 2. pull request 3. merge

Slide 23

Slide 23 text

all the pull requests

Slide 24

Slide 24 text

!! AVOID DOWNTIME !!

Slide 25

Slide 25 text

postgres <3 transactional ddl, concurrent indexes, cheap alter table add nullable columns

Slide 26

Slide 26 text

bidirectional compatibility

Slide 27

Slide 27 text

separation of state and connections

Slide 28

Slide 28 text

̿ CONTINUOUS TESTING ̀

Slide 29

Slide 29 text

sentry travis-ci.org test all the code

Slide 30

Slide 30 text

‘getsentry’ travis-ci.com test code relevant for us

Slide 31

Slide 31 text

̿ CONTINUOUS DELIVERY ̀

Slide 32

Slide 32 text

FREIGHT wait for travis > build > ship

Slide 33

Slide 33 text

bidirectional communication with the main slack channel

Slide 34

Slide 34 text

dev never matches prod :(

Slide 35

Slide 35 text

thus: fast rollbacks! (backwards + forwards compatibility)

Slide 36

Slide 36 text

̿ CODE STRUCTURE ̀

Slide 37

Slide 37 text

large systems are organisms

Slide 38

Slide 38 text

not all things will run the same code at the same time

Slide 39

Slide 39 text

data schema ~ code behavior

Slide 40

Slide 40 text

break up larger features

Slide 41

Slide 41 text

feature flag it! (we shipped some code to on-prem we backed out)

Slide 42

Slide 42 text

̿ MOVING PARTS ̀

Slide 43

Slide 43 text

keep dev basic: fewer parts

Slide 44

Slide 44 text

do not diverge dev from prod too much

Slide 45

Slide 45 text

virtual machines and docker are not an acceptable dev environment

Slide 46

Slide 46 text

̿ REPRODUCIBLE BUILDS ̀

Slide 47

Slide 47 text

pip freeze / npm shrinkwrap

Slide 48

Slide 48 text

nothing is more frustrating than a failed deploy because a dependency of a dependency of a dependency of a dependency pushed out a broken release

Slide 49

Slide 49 text

build once > ship to many

Slide 50

Slide 50 text

̿ MONITOR FAILURES ̀

Slide 51

Slide 51 text

associate failures to users

Slide 52

Slide 52 text

map support requests to failures

Slide 53

Slide 53 text

use sentry :-)

Slide 54

Slide 54 text

̿ FRIENDLY ROBOTS ̀

Slide 55

Slide 55 text

replace yourself!

Slide 56

Slide 56 text

bots and webhooks

Slide 57

Slide 57 text

github hooks

Slide 58

Slide 58 text

notify to communication hub

Slide 59

Slide 59 text

̿ BETTER CLIMATE ̀

Slide 60

Slide 60 text

the more robots, the better the integration, the smaller the fear of doing damage

Slide 61

Slide 61 text

If you can launch a feature on your first day of work that's motivating

Slide 62

Slide 62 text

also: happy customers

Slide 63

Slide 63 text

Q&A