Merge pull request #59535 from josh-hildred/tar_support

Add support for reading and writing backups as tar archives

contrib/libarchive-cmake/CMakeLists.txt

@@ -157,7 +157,7 @@ if (TARGET ch_contrib::zlib)
 endif()
 
 if (TARGET ch_contrib::zstd)
-    target_compile_definitions(_libarchive PUBLIC HAVE_ZSTD_H=1 HAVE_LIBZSTD=1)
+    target_compile_definitions(_libarchive PUBLIC HAVE_ZSTD_H=1 HAVE_LIBZSTD=1 HAVE_LIBZSTD_COMPRESSOR=1)
     target_link_libraries(_libarchive PRIVATE ch_contrib::zstd)
 endif()
 

docs/en/operations/backup.md

@@ -168,6 +168,28 @@ RESTORE TABLE test.table PARTITIONS '2', '3'
   FROM Disk('backups', 'filename.zip')
 ```
 
+### Backups as tar archives
+
+Backups can also be stored as tar archives. The functionality is the same as for zip, except that a password is not supported. 
+
+Write a backup as a tar:
+```
+BACKUP TABLE test.table TO Disk('backups', '1.tar')
+```
+
+Corresponding restore:
+```
+RESTORE TABLE test.table FROM Disk('backups', '1.tar')
+```
+
+To change the compression method, the correct file suffix should be appended to the backup name. I.E to compress the tar archive using gzip:
+```
+BACKUP TABLE test.table TO Disk('backups', '1.tar.gz')
+```
+
+The supported compression file suffixes are `tar.gz`, `.tgz` `tar.bz2`, `tar.lzma`, `.tar.zst`, `.tzst` and `.tar.xz`.
+
+
 ### Check the status of backups
 
 The backup command returns an `id` and `status`, and that `id` can be used to get the status of the backup.  This is very useful to check the progress of long ASYNC backups.  The example below shows a failure that happened when trying to overwrite an existing backup file:

src/Backups/BackupImpl.cpp

@@ -927,7 +927,7 @@ void BackupImpl::writeFile(const BackupFileInfo & info, BackupEntryPtr entry)
 
     const auto write_info_to_archive = [&](const auto & file_name)
     {
-        auto out = archive_writer->writeFile(file_name);
+        auto out = archive_writer->writeFile(file_name, info.size);
         auto read_buffer = entry->getReadBuffer(writer->getReadSettings());
         if (info.base_size != 0)
             read_buffer->seek(info.base_size, SEEK_SET);

src/IO/Archives/IArchiveWriter.h

@@ -22,6 +22,8 @@ class IArchiveWriter : public std::enable_shared_from_this<IArchiveWriter>, boos
     /// of the function `writeFile()` should be destroyed before next call of `writeFile()`.
     virtual std::unique_ptr<WriteBufferFromFileBase> writeFile(const String & filename) = 0;
 
+    virtual std::unique_ptr<WriteBufferFromFileBase> writeFile(const String & filename, size_t size) = 0;
+
     /// Returns true if there is an active instance of WriteBuffer returned by writeFile().
     /// This function should be used mostly for debugging purposes.
     virtual bool isWritingFile() const = 0;