Don't Panic!

A presentation at DevOpsDays London in September 2018 in London, UK by Euan Finlay

Slide 1

Slide 1

Don't Panic! How to Cope Now You're Responsible for Production Euan Finlay @efinlay24

Slide 2

Slide 2

Slide 3

Slide 3

https://theantimedia.com/bbc-accidentally-reports-the-death-of-queen-elizabeth-ii/

Slide 4

Slide 4

https://www.businessinsider.com/ft-publishes-incorrect-ecb-announcement-early-2015-12

Slide 5

Slide 5

https://www.businessinsider.com/ft-publishes-incorrect-ecb-announcement-early-2015-12

Slide 6

Slide 6

https://twitter.com/Ludo_Dufour/status/672401158653218816

Slide 7

Slide 7

https://www.ft.com/content/c01f37ec-99c1-11e5-987b-d6cdef1b205c

Slide 8

Slide 8

Slide 9

Slide 9

/usr/bin/whoami @efinlay24

Slide 10

Slide 10

/usr/bin/whodoiworkfor No such file or directory. @efinlay24

Slide 11

Slide 11

https://www.ft.com

Slide 12

Slide 12

You've just been told you're on call. (and you're mildly terrified) @efinlay24

Slide 13

Slide 13

Obligatory audience interaction. @efinlay24

Slide 14

Slide 14

Everyone feels the same way at the start. (I still do today) @efinlay24

Slide 15

Slide 15

How do you get to the point where you're more comfortable? @efinlay24

Slide 16

Slide 16

A tenuous link to A Christmas Carol.

Slide 17

Slide 17

The Ghost of Incidents... > Future Present Past

Slide 18

Slide 18

The Ghost of Incidents Future

Slide 19

Slide 19

Handling incidents is the same as any other skill. @efinlay24

Slide 20

Slide 20

Get comfortable with your alerts.

Slide 21

Slide 21

Get comfortable with your alerts. (and bin the rubbish ones)

Slide 22

Slide 22

Have a plan for when things break. @efinlay24

Slide 23

Slide 23

Keep your documentation up to date. @efinlay24

Slide 24

Slide 24

Practice regularly. @efinlay24

Slide 25

Slide 25

The one where we decommissioned all our production servers

Slide 26

Slide 26

Slide 27

Slide 27

Break things, and see what happens. Did your systems do what you expected? @efinlay24

Slide 28

Slide 28

The Planned Datacenter Disconnect

Slide 29

Slide 29

We got complacent, and stopped running datacenter failure tests... @efinlay24

Slide 30

Slide 30

The Unplanned Datacenter Disconnect

Slide 31

Slide 31

Have a central place for reporting changes and problems. @efinlay24

Slide 32

Slide 32

Slide 33

Slide 33

The Unplanned Datacenter Disconnect (Part II)

Slide 34

Slide 34

We should have followed our own advice. @efinlay24

Slide 35

Slide 35

We're not perfect. (but we always try to improve) @efinlay24

Slide 36

Slide 36

The Ghosts of Incidents... Future > Present Past

Slide 37

Slide 37

The Ghost of Incidents Present

Slide 38

Slide 38

Calm down, take a deep breath: it's (probably) ok. @efinlay24

Slide 39

Slide 39

Don't dive straight in. Go back to first principles. @efinlay24

Slide 40

Slide 40

What's the actual impact? @efinlay24

Slide 41

Slide 41

"All incidents are equal, but some incidents are more equal than others." George Orwell, probably @efinlay24

Slide 42

Slide 42

What's already been tried? @efinlay24

Slide 43

Slide 43

Slide 44

Slide 44

Is there definitely a problem? @efinlay24

Slide 45

Slide 45

Slide 46

Slide 46

What's the minimum viable solution? @efinlay24

Slide 47

Slide 47

Get it running before you get it fixed.

Slide 48

Slide 48

Check the basics first. @efinlay24

Slide 49

Slide 49

Slide 50

Slide 50

Slide 51

Slide 51

Don't be afraid to call for help. @efinlay24

Slide 52

Slide 52

The One Where a Manager Falls Through the Ceiling

Slide 53

Slide 53

The One Where a Director Falls Through the Ceiling

Slide 54

Slide 54

Slide 55

Slide 55

Communication is key. Especially to your customers. @efinlay24

Slide 56

Slide 56

Slide 57

Slide 57

Put someone in charge. @efinlay24

Slide 58

Slide 58

Slide 59

Slide 59

Create a temporary incident channel. @efinlay24

Slide 60

Slide 60

Slide 61

Slide 61

Slide 62

Slide 62

If you think you're over-communicating, it's probably just the right amount. @efinlay24

Slide 63

Slide 63

Tired people don't think good. @efinlay24

Slide 64

Slide 64

Slide 65

Slide 65

The one where we had to serve traffic from staging

Slide 66

Slide 66

Slide 67

Slide 67

It wasn't great, but it wasn't the end of the world. @efinlay24

Slide 68

Slide 68

The Ghosts of Incidents... Future Present > Past

Slide 69

Slide 69

The Ghost of Incidents Past

Slide 70

Slide 70

Congratulations! You survived. It probably wasn't that bad, was it? @efinlay24

Slide 71

Slide 71

Run a post-mortem with everyone involved. @efinlay24

Slide 72

Slide 72

Incident reports are important. @efinlay24

Slide 73

Slide 73

Slide 74

Slide 74

Prioritise follow-up actions. @efinlay24

Slide 75

Slide 75

https://blog.travis-ci.com/2018-04-03-incident-post-mortem

Slide 76

Slide 76

https://blog.travis-ci.com/2018-04-03-incident-post-mortem

Slide 77

Slide 77

Identify what can be done better next time.

Slide 78

Slide 78

The One with the Badly Named Database

Slide 79

Slide 79

Don't name your pre-production database: 'pprod' Seriously, who does that? @efinlay24

Slide 80

Slide 80

https://twitter.com/iamdevloper/status/1040171187601633280

Slide 81

Slide 81

Nearly the end. (don't clap yet) @efinlay24

Slide 82

Slide 82

Problems will always happen. (and that's ok) @efinlay24

Slide 83

Slide 83

The end. (please clap)

Slide 84

Slide 84

@efinlay24 euan.finlay@ft.com We're hiring! https://ft.com/dev/null https://aboutus.ft.com/en-gb/careers/ Image links: https://goo.gl/3DeojV