The previous chapter in this WhiteHat Security series discussed Codebase as the first step of the Twelve-Factor App and defined a security best practice approach for ensuring a secure source control system. Considering the importance of applying security in a modern DevOps world, this next chapter examines the security component of step two of the Twelve-Factor methodology. Here follows some actionable advice from the WhiteHat Security Addendum Checklist, which developers and ops engineers can follow during the SaaS build and operations stages ...
What if you discover a fatal error or an exploit in your app? What if your app is down during a crucial time? As a developer, how you react to a crisis can mean the difference between minor blip and an embarrassing or costly company blunder.
Your organization is depending on you to save the day. Here's a crisis management plan to get things right when they go wrong:
1. Stay calm and think clearly
This is obvious, but easier said than done. Staying calm will allow you to think more clearly. Working long hours under stress frequently leads to subpar code, and may be the reason why your app is down in the first place.
Don't underestimate the value of taking a walk, grabbing a snack or something else that changes your outlook to see the issue in a different light. And don't worry about wasting time, your brain will still be working on the issue in the background.
2. Revert to working code
If the appropriate solution might take a while to implement, or you have no idea how long the fix will take, roll back to a previous version of the code as a temporary measure. This is the last time you know your service was working, and it's a stable build.
Reverting to working code can provide the extra time you need to thoughtfully address the issue. When you're no longer rushed, you can gather the information you need to more effectively resolve the problem.
3. Monitor and alert stakeholders
You're a responsible developer, so hopefully you've set up some monitoring and alerting for your app. Before you ever find yourself in crisis mode, make sure you set up the right triggers and intervals for monitoring.
Once you receive the initial alert of a failure, notify the appropriate engineers of the outage so they can get started on a fix right away. Also notify other internal stakeholders so they're aware that their services might be impacted. As an alternative to notifications, you can set up a status page to inform those who rely on your app of updates on performance and availability.
4. Debug the issue
Now that you're set up for success, it's time to dig into the issue – starting with your logs. Logs are only as helpful as you make them. You should be logging the right activities with descriptive log statements. In addition to your existing log statements, add temporary log statements to guide the debugging process.
If you're still uncertain about the cause of the outage, focus on isolating the issue. Some code bases are structured in a way that is easy to see where the code is failing, but some dependencies and abstractions make it tricky to pin down the root culprit. If you're at a loss, you can try the debugging variation of a binary search by dividing and conquering to pin down where the code is failing in the most efficient manner.
Finally, don't work in a silo. You might benefit from talking through the problem or getting another perspective. Some developers like to pair program, rely on another teammate for rubber duck debugging, or use a literal rubber duck to slow down and articulate code line by line.
5. Push fixes with continuous deployment
When you're making code changes under pressure, you might be hacking together a solution. As a result, you might also be cutting corners and incurring technical debt in exchange for a quick turnaround.
Running an automated test suite guarantees consistent code coverage, and rigorous regression testing hedges against unintentionally affecting another dependency. Automating the build, testing and deployment process ensures you deliver patches in the fastest and most efficient manner possible.
6. Communicate changes to the team
Once you've patched the fix, communicate the status and diffs to the team. Just because your world came to a standstill with the problem doesn't mean the same happened for the rest of your team. They are continuing to work on their own features and issues.
Communicate what is necessary to keep them up to speed on the changes, and how it might impact what they're working on. Some teams keep track of an activity feed to stay up-to-date with the latest code changes, while others prefer to set up an integration with their preferred messaging platforms.