background

In fact, saying routine is not derogatory, but a wrong interpretation when we do not understand this, that this answer is compared with the Eight-part essay, can not directly reflect the technical value.

But, of course not, any kind of methodology or routines are based on the historical communication, successful product or project experience of checking, if you are not in accordance with the idea of success, I can do that, but in other scenarios may fail, because they do not constitute a closed loop, only the execution and some hash.

In the book “Self-Cultivation of Full-stack Engineers”, engineers should have many kinds of thinking, among which product thinking is very important. If the project experience we think can be described in terms of product thinking, it will get twice the result with half the effort and be easier to implement in the relevant ecology of big factories. In fact, most small factories follow this pattern after iteration normalization.

What if the bug rate is higher online

If you come up here, the answer is:

  • 1. Categorize, analyze and solve the recurrence process
  • 2. Reverse check problems from the research and development stage
  • 3. After comparative analysis of different environments, why did other environments not recur
  • 4 Core functions for internal testing, gray release
  • 5. Determine the liability mechanism
  • 6 Determine the rollback mechanism
  • 7,,,

Chances are you’re dead on this one (assuming it’s already two or three), and you’re probably wondering why? This is not a very normal and comprehensive set of ideas, the problem is not your plan, but your set is wrong. You start planning without knowing the details of the problem, and you may end up doing nothing and not getting paid.

So the second and third interviews ask you such questions, the subtext is to ask you how to deal with this matter, rather than how to do purely technical solutions, that is very late work.

One possible way to think about it is this

What is the definition of high bug rate online

Including but not limited to:

  • Data collection source
  • Whether it is a bug
  • High standards of definition
  • Which ones are higher priority
  • Summary of bugs

Talk about expectations

The feedback side of the question should propose what is the expected result of this phenomenon, including but not limited to

  • To what extent are expectations lowered
  • What is the expected bug resolution rate and the lifetime of the fix
  • How do you expect to get specific error information, user information and so on in the first time when a bug occurs
  • Expect to know the side effects of the bug
  • Expect the r&d team to pay attention to this matter, and the proportion of invested personnel

Talk about programs and input-output ratios

Knowing what the other party wants, the r&d team can then make corresponding solutions based on the existing experience and problems found, including but not limited to

  • How to detect problems earlier
  • How to reproduce the problem
  • How to balance the impact and damage of bugs
  • How to solve the problem and shorten the problem cycle
  • How do you roll back functionality when needed
  • How to quickly roll back code
  • Project accountability mechanism, person in charge
  • In the major function of the release of internal testing, gray release, etc

Provide a solution for each problem, and the cost, and reach some agreements with the business side, which solution to use, how much cost to invest, and what effect to achieve.

Abstract, prioritize, and accumulate systematic solutions

All the problems mentioned above are very large and systematic. Technically speaking, it is impossible to solve all of them overnight. Therefore, the most conventional solution must be to solve the most important problems, the problems with the highest input-output ratio.

Develop a standard solution for the highest priority and most cost-effective set of solutions, processes, indicators, and effects.

So when you see something like this, you use this abstract scheme; If there are upgrade issues or extension issues, improve the solution;

As you have more and more solutions, they form a large system of closely interrelated positive influences, rather than a pure technology point.

Solve the problem and come up with a concrete plan

For a specific solution direction, list specific technical solutions, such as how to find the problem earlier. It can be as follows:

  • 1 User behavior monitoring — user behavior buried point
  • 2 Application Exception monitoring: Monitors exceptions
  • 3 Interface data anomaly detection – Monitors back-end anomalies
  • Regular user behavior analysis — user research, user portrait

Rematch, expand the results

The business side delivers the results of the review done well and badly.

If the results are good, then expand the scope of use, maximize the value of the scheme.

summary

The above is my perfect closed-loop logic when the interviewer asks how to solve the problem of “high bug rate online”. In the subsequent interview process and work process, I will try my best to promote the corresponding work in accordance with this way of thinking.