Several common problems of SQL server synchronization database without public IP across network segments

Problem description

host 1： Publisher

Aliyun server -- There are public ip

host 2： Subscribers

The notebook -- No public network ip

Small amount of data , Mainly familiar with the operation process of publish and subscribe .

host 2 Just as a host 1 Local backup of , After you request to modify the data on the ECS , Can pass sql server The publish and subscribe function of will synchronize the local data .

There is no step-by-step introduction , Step by step , You can read the following

problem 1

Ali cloud, sql server After configuration , Can't use local sql server Client remote login .

Check the following

1. The firewall in alicloud's console , Open Alibaba cloud's 1433 port （ default sql server Access port ）

That's important , There is no mention of ！！！

2. We need to take the sql manager Medium sql server In service sql server Agent on （ It was originally disabled ）

3. Remote server to open sql server User name password login method , And set up the proxy account and password

To this step , It should be in the notebook sql server adopt ip Address , And just set up the proxy account and password to log in to the ECS .

--------------------------------------------------------------------------------------

problem 2

Publish and subscribe process

First in the cloud sql server Client's

Copy --- Release --- New release

1. Choose which way to publish

In this regard, you can query the differences among the four publications

it is to be noted that ： When the transaction is released , The published table should have a primary key

2. How to connect to the publisher

On the Internet, it is generally recommended to create a new user name in both the cloud and the local , The code is the same windows Account , Then here we choose windows Run under account .

I'm going to use sql server Login name

Note here --- The login name is sql server Login name of （ It's usually sa the ）

-------------------------

That's the publisher

-------------------------

Subscribers

1. Use the alias

When choosing a publisher in this step , Use the actual server name （ Use windows The name of the login method ）, Out of commission ip Address .

Here, add the publisher alias to the subscriber , See below for details

https://blog.csdn.net/u010457730/article/details/97276313

2. Push subscription or pull subscription

Here I have a question I don't understand , I'll talk about it later .

3. Connect to the distributor

Note here , To fill in the sql server Login name .

It's the login ,sa that

Login is not a server name , Remember this .

At this point , Check the synchronization status of subscribers

See if it's ticked ---- This means that the previous settings are correct , Subscribers can access the publisher .

I filled in the wrong user name at the beginning , When checking the synchronization status, an error is still reported .

--------------------------------------------------------------------------------

Next , Let's see if the subscriber's data is synchronized , If not synchronized , Look at the job history of the subscriber

Maybe it's still a mistake

But what is this misinformation ？？？

“ Replication agent is having problems . For more information , See previous job step history messages or Replication Monitor ”

Originally, we need to find the specific error information at the publisher ...

Start the Replication Monitor at the publisher

After selecting a specific release on the left , Double click the wrong subscription in all subscriptions on the right , View specific error reports

The truth --

Due to an operating system error 3, The process could not read the file D:\XXXX\X.pre

The reason is -- This path is where the snapshot file is stored at the publisher , But because it's not a network folder , Can't be visited .

Here's a simple solution

Follow this example

https://www.cnblogs.com/mrray/archive/2011/03/17/1987123.html

Directly put the snapshot folder of the publisher -- The name is repldata Copy of to a file location on the subscriber side

Modify the snapshot location at the subscriber side

Copy -- local subscription --- Specific subscriptions --- Right click --- attribute

Change the snapshot folder location to the local location .

It's amazing , It only needs to be modified once .

To this step , My problem is solved , Here are some questions and thoughts

1. Many holes , But look for more information , It can always be solved . This confidence is exercised in solving problems over and over again .

2. Lack of theoretical basis , Some configuration operations are prone to problems . For example, I don't know the specific mechanism of publish and subscribe , It's hard to think of the problem that the snapshot folder can't be accessed .

3. My situation is quite special , One has a public network ip, None of them . What I saw on the Internet , Usually it's a LAN , Or both hosts have a public network ip Of （ This means that the publisher , Subscribers can use only ip Address visit each other ）. So I've always suspected that my network structure can't use publish and subscribe in principle （ Because the publisher can't “ find ” Subscribers ）. This is also my choice to run the agent from the subscriber in the configuration , Because subscribers can definitely “ find ” There are public ip Of the publisher . This problem needs to be studied in detail .

4. These four releases - What's the difference in the way you subscribe ？ What is the principle of snapshot folder ？ These are two questions that need to be clarified .

5.sql server Why there are so many pits for this common function ... Maybe I'm too busy . But at least this problem is solved , I've grown a little bit .