Preface
The host is undergraduate slag 211, master's degree in Chinese, all majors are SE, there is no internship experience when the school recruits, and he has been working in a large factory for more than two years, doing back-end development work. There is no high-concurrency case or high-traffic project. , And there is no scenario where some middleware or nosql must be introduced to solve business problems, referred to as "three noes", you can only try to draw generals from the short ones in your resume and write some content that may have some value. In the process of social recruitment, I took offers from Meituan, Huolala, Weizhong, Ali, and byte, and finally chose byte.
The biggest feeling about social recruitment
When the school recruited, because I did not participate in the internship, they are basically all toy projects in the laboratory. The business complexity and technical complexity may be O(1) level. The most interviews are basically basic knowledge, framework principles, etc. There is nothing to say. Just talk about the project casually.
When social recruitment, if your field or direction is not in line with the job interview, the difficulty of social recruitment interview will basically rise sharply, relying on technical strength; if you are in a similar field, then there may be many projects involved , The technology is not that difficult, if you happen to be a domain module they need, then you will have a greater chance of success in the interview.
Social recruitment interview project is very important, not only the technical complexity of your project itself, but also the business complexity, what role you play in the project, what problems you have encountered, is it a technical bottleneck problem, or a business problem, how to solve it Yes, these few questions are very important
Many classmates, including myself, didn’t have a clue at the beginning. I read many interviews and said that they should be so familiar with their resumes and their own projects. So, what is considered as so-so, just memorize the content of the resume. ?
The following is a summary of some of the issues. You must be aware that you must be self-consistent in your logic, not inconsistent, and think clearly about the causes and consequences. You can also not fill up all the questions by yourself, you can deliberately miss the questions and wait for the interviewer to ask questions (interview routines, self-directed question direction). In short, otherwise you may be asked a question that is not in this scope, and you don't know how to answer it, which will give the interviewer a bad impression (why don't you understand your own project?)
Finally, summarize the most important questions in the interview:
- 1 Why is your project designed like this?
- 2 What kind of problems does this design deal with? What are the benefits of your design like this, and what benefits have you gained?
- 3 Is there any bottleneck in this design? What's the problem? What are the disadvantages of this design? Poorly considered?
- 4 How to solve the problem? How to optimize for bad places?
If you come from a big company, if you ask three questions, or two don’t know, or you get asked once during the interview, you get stuck or think poorly because you have not prepared your questions, then the interview will be greatly reduced. Dachang is not a halo, but a problem. (Big factories should be okay, how come you are not good enough?) The
interview must be confident. Originally, applying for a job is a two-way selection process. It is not that you do not enter the interviewing company. Don't be too low just because you are the interviewer. It is wrong to beg the interviewer to give you a job. A good interview should be for you to confidently express your own understanding. Even if you don’t, don’t be embarrassed. You don’t need to hesitate or talk nonsense. This is a taboo. No or no, you can answer dashingly, it doesn’t matter, a good company will not keep asking questions you don’t know, and it doesn’t make sense.
The host agency recruits interview experience
In fact, social recruitment is still very unfriendly to the host who has to do his job well during the day and night and respond to demand. If you are busy working overtime at night, it will be even more miserable. The host started to review in July and August. In fact, the progress of the review is very slow. First, try to spare as much time as possible to charge at night. If you get home from get off work at 10 o'clock, after a short break, it will start at 10:20 or 10:30. Charging, it’s best to review for a little more than an hour, almost see 11:30, ready to wash and sleep. But if the commute time after get off work is also very long, I feel that I can charge for a while on the road.
On weekends, it is even more important to arrange time. This is a rare whole piece of time. First of all, you must make an outline, which modules are included in the content to be reviewed, and which modules to review this weekend. Then adjust the time with your girlfriend or family, stabilize your relationship, and explain what you are doing. Don't just because this kind of thing affects your relationship and your review progress.
Regarding the resume, it is difficult for the host to write a resume. For the technology stack, he is afraid to write proficiently. Those who don't know the principles in particular don't dare to write more, so the technology stack looks relatively weak. For the resume project, there are not many places to shine, and the things that I have accumulated and summarized are also lacking. But some skills, such as which projects you are responsible for, which modules are responsible for, what functions are implemented, how much performance you have improved, may put a number to better reflect your value, for example: through the optimization of xxx, query speed can be increased by 100%;
Through the optimization of xxxx, the non-perceptual expansion of xxx seconds is realized; through the optimization of xxx, the processing speed is increased by 200%, etc.
Regarding the interview arrangement, the host started to upload his resume on the boss in September, and then there are many greetings, so you need to filter out the positions you are interested in. Personal suggestion, you may need to make a good job in some positions. For example, if you want to join Ali, then you cannot interview Ali at the beginning. Of course, the big guys skip it, because the interview experience is also a gradually proficient process, and the status is also improved. The process may not be fair to the previous company, but it is indeed an effective method. However, it is a painful process for applicants who are recruited by the agency, who are on the job and may have more work tasks. Most of the interview time is arranged in the evening, and I set aside an hour for the meeting room. Maybe I need to continue working after the interview. Or it’s even more nervous to catch up with the release. You may still face the embarrassing situation of needing you to solve the problem at the work station at this time, and then you go to the interview, and you are still worried, hey, it is hard to say.
Therefore, the interview arrangement cannot have an interview every night. The interview process of each company is slow or fast. Then you need to coordinate how many new companies are invested each week. There may be several interviews in the original interview process this week. . The host's personal suggestion may be 2 in a week. This varies from person to person. It is very important to remember the summary and problem analysis after each interview. Don’t think that the interview is over. If you feel that the interview is not good, you will be discouraged. The interview questions are where you need to improve. This is a quick and effective way to improve yourself.
The basic time mileage is preparations in July and August, and investment in September. The initial stage is to accumulate experience. At the end of September and October, you will start to invest in your own companies. Near the end of October, November is the final stage, and then the selection process.
Face Sutra
Maybe some friends are in a hurry. Having said so much, what about the dry goods I want? Okay, I will prepare it for you. The chapter includes a lot of knowledge points, including basic knowledge, Java collections, JVM, multi-threaded concurrency, spring principles, microservices, Netty and RPC, Kafka, diary, design patterns, Java algorithms, databases, Zookeeper, distributed cache , Data structure, etc. Friends in need can click here to remark csdn to download directly, I hope it will be helpful to you, and I am sincerely waiting for you to get it!
One side of ape tutoring
Interview format: Niuke.com
Algorithm: Use linked list to simulate
the difference between oracle and mysql, isolation level
How to locate slow query? How to optimize? What kind of case have you encountered and how did you solve it?
The project uses sub-library and sub-table, will sub-library definitely improve performance? What is hot and cold data? What has been optimized? What if there is a surge in data? Is there any way to expand the capacity? How to expand without perception? How to achieve real-time data consistency?
The thread pool, core parameters, and waiting queue are full. Will the task be executed immediately?
The difference between spring boot and spring, what is spring boot used for? What is the biggest role? What is the starter?
hashmap1.7 and 1.8? Optimization point? Why is the number of red-black trees 8? Why is the degradation 6?
Two, Meituan
one side
Interview format: Niuke.com
Project introduction, difficulties, how did they meet? cause? How to solve? Why do you want to divide the table like this? How to solve cross-database join? How to solve the sudden increase in data volume?
How to solve online problems? What about cpu hurricane? What to do with frequent minor gc? What is the possible cause? How to avoid it?
How to achieve the database isolation level? Current read, snapshot read? MVCC?
Algorithm: the sum of two numbers
Two sides
Interview format: Niuke.com.
Project difficulties, how to solve them?
The design of sub-database and sub-table?
Have there been inconsistencies in distributed transactions? why? How to deal with it? Is there any way to avoid it? How to monitor? How to deal with the monitoring? When is manual access required
? How to write code elegantly? What code counts as elegant? What code is the norm? What is your code specification?
How to perform code review?
JVM model, when is the pc counter empty?
Algorithm: ZigZag traverses the binary tree
Three sides
Interview format:
Niuke.com What do you think about high traffic and high concurrency? Do you have any projects? Why not? What if so?
What do you think of your work since you started?
How can you get a good performance? What do you think of your performance?
What do you think about middleware? Which middleware have you used? To solve what problem?
How do you understand redis? What scene is suitable? What do you use it for? What problem is solved?
What do you think of our team?
What is your plan for yourself?
Why choose to leave at this time?
Three, the side of the micro public
Interview format: Niuke.com
. For the question, one question is the atomic integer class implemented by yourself, why is it not atomic; the second question is how to use the set functions of the spring bean life cycle? ; The third question is the problem that may be involved in the transfer of bank account A to B, deadlock? Affairs? database? Concurrency? Strong consistency? Have a talk
Four, the cool side
Interview format: telephone
project introduction, sub-database and sub-table
public private protected difference, put on the class, method, interface, attribute?
What is unmodified?
The difference between abstract class and interface, abstract class abstract method
function execution time is uncertain, it takes a certain time to return? How to achieve it?
When will the hashmap red-black tree be expanded? How does the expansion work? How does the ring happen? Infinite loop? 1.8 optimization?
How does concurrenthashmap achieve thread safety? Is it 100% safe? What is the difference between 1.7 and 1.8? Why do you do this?
Can I not set the primary key under mysql innodb? Can the primary key be empty? How many are allowed? What is the difference between a primary key and a non-primary key? Index storage format?
Joint index failure problem? Index invalidation problem?
What does the .java file contain? What does the .class file include? Java object storage structure
Five, praise
one side
Interview format:
What has been done to optimize the phone project?
Thread pool, parameters, how to set? callable and runnable? Communication between threads?
mysql index, covering index? Back to the table? What does the B+ tree leaf node store? Why not use AVL tree?
Database lock, optimistic, pessimistic, record lock? next-key lock?
When Hashmap 1.8 optimizes
gc, can young gc only have two areas? What's the problem?
Two sides
Interview format: WeChat video.
Project introduction. The
most difficult point.
High-load, high-traffic business processing.
How to achieve data consistency?
How to reduce manual intervention?
Six, headlines
one side
Interview format: Niuke.com
Special note: Toutiao is the only company that requires your code to be AC under Niuke.com’s IDE. The direct interviewer may give you a test sample for you to test, usually 20 minutes
Algorithm: Find change
Why is Redis so fast? Multiplexing io model? Compared with multithreading? select epoll model?
How does epoll work? What storage structures are used? When is the ready event? Where is the callback function registered? How does epoll use the steps? What did epoll wait do? How to process a get data request to the redis server?
What is the data structure of zset? Why use skip tables? What is the process of inserting a data? What is the process of getting a data? How is the multi-level index updated? How to determine the level of a data? What is the respective time complexity? What do you think is there to optimize?
How is the high performance of Redis achieved? Multiple masters and multiple slaves? One master, multiple slaves? What is the process when each key is inserted? What is master-slave replication like? How strong consistency? When is it not suitable? The Lord is dead, what will the respective elections look like? How to complete the main/standby switch? If the master hangs and never replicates completely, what will happen? Under what circumstances can such a grouping method be used? Under what circumstances is not suitable?
The raft protocol paxos algorithm
http status code 300 400 500 499
dos ddos drdos how to avoid it? How to prevent it? How did you find it? What features of TCP are used? How to improve TCP can be prevented? What if the server cannot handle the request? Does the maximum number of connections need to be set? How to set it up?
What does the Thrift RPC process look like? How do heterogeneous systems complete communication? What are the advantages and disadvantages compared with http? Do you know grpc?
https http1.1 2.0 3.0
What if hashmap really needs a large amount of data to operate? How to optimize?
Two sides
Interview format:
What will happen to Niuke.com's Linux fork subprocess if it forks in a thread? If there is a total of 4g of memory, and a process has 10 threads, can it be fork out in the thread?
There are several states of TCP connection between server and client.
How many TCP connections can linux as the client establish at most? As the server side?
When establishing a tcp connection, how do I block when connecting, then how can I do other things without blocking?
How does linux monitor the operation of the cpu? Port connection situation? tcp connection situation? How much memory is used?
How to handle multiple epoll requests? When is fd readable and writable? Does the epoll model use any queues to wait for? Does the epoll model process one request? Can you write some pseudo code for the processing flow?
How does Redis handle requests? What is the qps of a single machine? If you keep using get, what do you think is the limit of qps? How can we break through the peak qps? How to handle requests with single thread? How to solve the problem of peak qps of a single machine?
How to implement redis distributed lock? How to understand? Can I not use redis? What's the problem? Have there been any problems in production? Why does it appear? How to optimize? Have you considered zk? Stand-alone redis distributed lock? Distributed lock in cluster state? Distributed lock under master-slave? What's the situation? How does redlock understand?
How is the system highly available? What can multiple deployment machines achieve? What are the disadvantages? What are the advantages? How to achieve high availability? Can multiple deployment machines achieve high availability? What is the fusing mechanism? How to achieve it?
Three sides
Interview format:
Niuke.com Algorithm: Topk problem of summation between two ordered arrays.
How to do qps aggregation in sub-database and sub-table order of magnitude system?
How to do expansion?
Redis elimination strategy? What happens when the maximum memory limit is exceeded? When will redis be slow?
What problems can redis persistence mechanism nosql es solve? What about hbase?
Strong consistency? Weak consistency? For what occasion? ***How to achieve strong consistency on the web? How to design the cache?
Seven, cargo pull
one side
Interview format: basic data types of zoom
hashmap
sync and reentrantlock
redis, how to realize the bottom layer? Jump table? Why is single thread faster? What are the designs? How does a single thread handle multiple requests? What might cause redis to be slow? How does redis achieve high availability? Which commands might be slow?
Online oom situation? How to troubleshoot?
mysql mvcc double write mechanism? What is the order of binlog and redo log? How to deal with an unsuccessful writing? Current read and snapshot read? Transaction isolation level? Online SQL optimization experience?
Distributed transaction? What methods have been used? What else do you know? How to achieve it? Have you ever thought of implementing it yourself? Which pits to avoid?
How to achieve cross-database aggregation? Paging? which page? Next page?
Eight, Ali
one side
Interview format: telephone.
Special note: Ali interviewers will introduce their team before the interview. The outline of this interview is
how professional JVM interacts with the operating system.
What is garbage collection? How to recycle? Garbage collection algorithm? Garbage collector?
How does sync and reentrcntlock
sub-database sub-table aggregation query limit realize the realization of top to
expand capacity without stopping? Sub-table to avoid hot and cold? Expand the library without stopping the machine? Expand tables without stopping? Cross-database transactions?
Why is the thread pool parameter not fixed?
How to realize the high availability of the master-slave cluster of redis elimination strategy?
Distributed lock
zk realizes distributed lock?
Algorithmic question: write a few threads to send to his mailbox and print playing cards of different suits in turn from small to large
Two sides
Interview format:
Why is the telephone sub-database and sub-table designed like this? How to do data growth? How to expand? What to do if the data is uneven? How to separate hot and cold data? How to do aggregation? How to do cross-database aggregation and how to do query? How to do cross-database paging?
Group mode on mysql? One master, multiple slaves? why is it like this? How to ensure strong consistency? To solve the separation of read and write? Is it for one master and multiple backups? What if the main library crashes? From the library?
How to do distributed transactions? What principle? How did it happen? Have there been any transaction inconsistencies? why? How to solve it?
What to do with the surge in requests for access? How to relieve stress?
How is redis distributed lock implemented? Is it reliable? What's the problem? Can redlock solve it? why? How to optimize?
Why use redis?
How to deal with the soaring cpu? How to deal with load surge? Under what circumstances may encounter the above situation?
How is full gc formed? What situation would produce such a situation? How to troubleshoot online appearances?
Nine, Baidu
One side was over, and then I didn't ask hr about it twice, and hr told me that I would not recruit, embarrassing, and I forgot the interview question. . .
Ten, fast hands
Interview format:
How to do Niuke.com's top 100?
How to design the mobile terminal when browsing the web?
How to cache the favorite list?
How can everyone see a gift in the live broadcast room?
Designing a recently viewed work, how does the database and the cache work together? When to update the database and when to update the cache?
Algorithm problem: add two large numbers
Write at the end
The interviews are in no time order. It is possible that several of the interviews have not been written or finished. Because some interview interviews were forgotten at the time, they were forgotten. I attach my interview information and interview summary below. The content is roughly as follows, I hope it will be helpful to you:
Basically, it is similar to the original poster's idea. Anyway, the previous ones are not very good, and the latter may be better with experience! The above information is obtained free of charge, friends in need can click here to remark csdn to download by themselves, I hope it will be helpful to you! In addition, a friendly reminder, make a decisive investment when you see the position. It is very likely that the position is very hot. You always want to go to the face when you are well prepared, and the position may be gone. In addition, most companies need algorithmic questions, most of which come from simple and medium difficulty. I personally think that it is easy.