The migration tool supports checkpoint restart. If the upload of a large file is interrupted due to the tool error or service failure, you can run the tool again and restart the upload from the checkpoint.
No. All successfully uploaded files will be recorded in the db
directory. Before the migration tool runs, the db
directory will be scanned first, and the files recorded in it will not be uploaded again. For more information, see Migration mechanism and process.
Check whether your key, bucket, and region information is correct and ensure that you have the operation permission. If you're using a sub-account, it needs to be authorized by the root account. If you migrate data from a local file system or other cloud storage, write/read permission on the bucket is required. For the "Bucket copy" operation, read permission on the source bucket is also required.
This error occurs when the data download from other cloud storage times out due to insufficient bandwidth. For example, when you migrate overseas data from AWS to COS, "read time out" may occur due to network latency caused by insufficient bandwidth. To solve this problem, you can increase the network bandwidth and test download speed with wget
before migration.
This error occurs when frequency control is triggered. A limit of 30,000 QPS is imposed on an account in COS. We recommend you decrease the concurrency for small files in configuration. Then run the tool again to resume the migration.
Check whether your key, bucket, and region information is correct.
The possible reason is that the tool uses RocksDB which requires 64-bit JDK. Check whether your JDK version is x64.
In a Windows environment, the tool needs to be compiled in Microsoft Visual Studio 2015. In case of the above error message, you need to install Visual C++ Redistributable for Visual Studio 2015.
Modify the file src/main/resources/log4j.properties
by replacing the value of log4j.rootLogger with the log level, such as DEBUG, INFO, and ERROR.
/tmp/librocksdbjnixxx.so: ELF file OS ABI invalid
error is reported in the Linux environment?IFUNC needs to be supported on Linux and the binutils version in the running environment should be later than 2.20.
error.log
?The exception information in error.log
is similar to:
2022-XX-XX XX:XX:XX [ERROR] [main:xxx] [com.qcloud.cos_migrate_tool.task.MigrateLocalTaskExecutor:] [MigrateLocalTaskExecutor.java:183]
walk file tree error
java.nio.file.FileSystemLoopException: /dataseal/xx1/file1
at java.nio.file.FileTreeWalker.visit(FileTreeWalker.java:294)
at java.nio.file.FileTreeWalker.next(FileTreeWalker.java:372)
at java.nio.file.Files.walkFileTree(Files.java:2706)
at com.qcloud.cos_migrate_tool.task.MigrateLocalTaskExecutor.buildTask(MigrateLocalTaskExecutor.java:176)
at com.qcloud.cos_migrate_tool.task.TaskExecutor.run(TaskExecutor.java:244)
at com.qcloud.cos_migrate_tool.app.App.main(App.java:135)
The reason is that the file "/dataseal/xx1/file1" to be migrated may be a soft link pointing to a resource in its parent directory. You can check with the following command:
[root@TENCENT64 /dataseal/cos_migrate_tool_v5-master/log]# ll /dataseal/xx1/file1
lrwxrwxrwx 1 xx xx xx xx x xxxx /dataseal/xx1/file1 -> ../xx1/
As shown above, the soft link file "/dataseal/xx1/file1" points to "/dataseal/xx1/" in its parent directory, which causes an infinite loop in traversal. Therefore, the migration task is automatically terminated.
We recommend you delete such files in advance (note: the method of excluding such files in the configuration item excludes
is invalid).
If other problems occur, try to run the migration tool again. If the problem persists, compress the configuration (with the key hidden) as well as the log directory and contact us.
Was this page helpful?