About retrieving password by email

About retrieving the password by email, what is the general idea to achieve it?
Thank you

Reply content:

About retrieving the password by email, what is the general idea to achieve it?
Thank you

Your question is very brief, but covers a wide range.

There will be many safety issues involved, and there may be unsafe factors if you are not careful.

First of all, give you the textbook-level OWASP password recovery design guide:
https://www.owasp.org/index.php/Forgot_Password_Cheat_Sheet

Then I will give you three dark cloud articles. Use other people’s mistakes as a reference to see if you have similar unsafe designs:

1. Summary of password retrieval logic vulnerabilities:
   http://drops.wooyun.org/web/5048

2. Possible problems with the password retrieval function
   http://drops.wooyun.org/papers/287

3. Possible problems with the password retrieval function (supplementary)
   http://drops.wooyun.org/web/3295

Let me tell you my thoughts first:
1. The user enters the email address and goes to the database to check whether the email address exists
2. If it exists, a routing connection is immediately generated, such as www.a.com/lost/ passport/key/bac.html
3.key=>bac bac is a unique hash value generated by some kind of calculation. It is placed in a cache system such as redis and is valid within 24 hours, or 2 hours
, so Send this connection to the mailbox, the user clicks the link and executes password retrieval
4. When retrieving the password, the first step is to verify whether the key hash value exists in the redis cache, and then execute update to reset the password. That’s it.

Generally, there is a time limit for retrieving passwords, so time is indispensable, and the users who need to change must also be included.
Just find a reversible encryption function with a key to process these two pieces of information to generate a token.

One of my ideas does not require the use of redis and database, and there is no need to independently store the hash of the user's password retrieval.

The method someone mentioned above is that in large-scale business, you have to build several independent redis servers, which is a bit expensive.

My idea is

1. First use `
   md5 (user email, current timestamp, private key) user email clear text, current timestamp clear text

<code>点开这个链接之后先判断时间戳合当前时间之间是否超时。没超时则再次用```
md5(用户邮箱，当前时间戳，私钥)</code>

Since the private key in the entire process is in the server-side script, as long as it is not leaked, the system is safe There is no problem with the whole process in theory.

Everyone above said it better. The author can summarize it