Put official documents and migration-related FAQs first
I used the tools in the online migration to migrate from Tencent Cloud to Tencent Cloud.
Compatibility and tool configuration instructions say that tool migration has public network migration mode and intranet migration mode
Public network migration mode
If both your source host and target cloud server have public network access capabilities, you can use the public network migration mode for migration.
Intranet migration mode
If your source host or target cloud server is in a certain intranet or VPC, and the source host cannot directly establish a connection with the target cloud server through the Internet, you can use the intranet migration mode of the tool to migrate. The intranet migration mode needs to establish a connection channel between the source host and the target cloud server by using methods such as VPC peering connection, VPN connection, cloud networking, or dedicated line access.
Please back up important data manually before migration!
first try
I have a public network ip, I thought I could only use the public network migration mode, but the documentation and FAQs say that the migration speed is related to the number of files and the bandwidth of the server's external network. I don't have many files, and both bandwidths are 1Mbps. According to the steps in the document, the migration mode using the public network is very slow, and it took more than ten hours. Then I want to cancel the migration.
In the part of this page工具运行参数说明 , the parameter is 工具迁移that --cleanthe target cloud server is forced to exit the migration mode and clean up the site. But I --cleandid not successfully stop the migration with the parameters, and reported an error Other go2tencentcloud are running., and then I forced kill -9all tencentthe processes with it to end, and then executed --cleanthe parameters and still reported an error (as shown in Figure 1, the kill command has been executed in the screenshot). As an aside, how does xshell look like this, see my article .
Then I tried other parameters, and they all worked normally, but it --cleandidn't work. Then after a few hours, I suddenly thought, try --helpthe introduction in the parameter result, the result is printed in Figure 2, the parameters here are all -, not in the document --, I don’t know the difference between them, if anyone knows, please tell I tried -cleanthe parameters again and successfully stopped the migration.
The reason for the success, I don't know is because of waiting for hours, the migration process stopped; or the parameter should be used -cleaninstead of the document --clean.
second attempt
Just when I encountered --cleanparameters that could not stop the migration, I consulted the online customer service, and the customer service said to let me contact the QQ group . At the beginning 工具迁移, I chose it because 工具迁移I didn't need to apply for joining the group, but 控制台迁移I needed to apply for joining the group (as of writing this article), and now I have to join the group.
No one in the group solved --cleanthe problem of not being able to stop the migration for me. -cleanAfter I successfully stopped the migration, I told my method in the group. Then the official replied to my picture 1:
At this time, the migration tool process is running. If you need to cancel the migration, you need to kill the migration tool process, and then execute –clean, so that the current migration is stopped and you can migrate again
Then I said that I had executed before this kill -9, and he said:
Cannot kill -9
kill -9 will cause the tool process to fail to exit normally
But it didn't say how to make the migration tool exit normally.
Now that I have stopped it (although I don't know the real reason), then I asked, I am migrating from Tencent Cloud to Tencent Cloud, can't I hurry up? He said:
For destination machines that are billed as you go, the bandwidth of the destination machine can be adjusted during migration.
Tencent Cloud machines can be migrated directly through the internal network without going through the public network, and are not subject to public network bandwidth and fee restrictions.
What I'm waiting for is his second sentence! Going back to the introduction of the intranet migration mode at the beginning of the article, it did not say that it can only be used on intranet machines (without public network ip). After I expressed my doubts, he said:
Documentation has no such restrictions
And gave me an introduction to Tencent Cloud intranet services .
That is to say, I have a public network ip, and I can also use the intranet migration mode, which is much faster. In the intranet service, it is said that the bottom layer is 10G/Gigabit interconnection.
migration result
Scenario 2 of the intranet migration mode I use, I have few files, and the migration is completed in 8 minutes. Scenarios are introduced in the section Compatibility and Tool Configuration Instructions .支持的迁移模式工具迁移
I migrated from CentOS8 to CentOS7 ( the document says that the system version must be the same, otherwise there may be problems ). After the migration, the login password of the destination server (CentOS7) becomes the login password of the source server (CentOS8) . The previous pagoda address Changing the ip to the ip of the destination server can also log in, the account password remains unchanged, and the pagoda data of the source server is still logged in. The original docker container in the pagoda cannot be started, and it has not been tested whether the database installed through the pagoda can be connected.
If the migration is normal, it seems that the source server has been mirrored and all data has been synchronized, but the system version of the destination server will not become the system version of the source server .
Next, we need to solve the problem of the docker container. Since the container cannot be started (the screenshot was not taken at the time), thinking that even if the docker is reinstalled, it may still not be able to start, so I uninstalled the pagoda and reported an error after reinstalling. There is a link download failure, 404, link There are centos/8words in it (no screenshots at the time), similar to the errors reported in these two articles ( refer to link 1, refer to link 2), it seems that it may be due to the system version, which is actually CentOS7, but the source server is CentOS8 when migrating, so I will reinstall up the server.