如何创建一个“适合”子目录中文件的块设备？

Question

I need to create and format a block device and copy all the files in a subdirectory into it. Ideally there will be just enough space on the formatted block device to fit the files. It doesn't matter if it's not a precise fit but it should be fairly close - within tens of megabytes ideally.

But its not as easy as it sounds. First I need to measure the size of the files in the subdirectory, and then need to create a block device.

But the block device cannot be the same size as the files because some space will be taken up by formatting information I assume. Also some space will be taken up because presumably files don't necessarily use all the available space within each disk block.

I don't think it matters much what sort of block device I am creating but in case you think it is important, I am creating an EBS volume on Amazon EC2.

So does anyone have any comments on how best to do this?

Here's my Python code for calculating size of the directory:

def get_directory_size(start_path=None):
    print(start_path)
    total_size = 0
    for dirpath, dirnames, filenames in os.walk(start_path):
        for f in filenames + dirnames:
            fp = os.path.join(dirpath, f)
            if os.path.islink(fp):
                # ignore symbolic links
                continue
            total_size += os.path.getsize(fp)
    return total_size

Answer 1

If you are using a filesystem in the ext family (eg, ext3 or ext4), you can start with a filesystem that's larger than needed, then use resize2fs -M to resize the filesystem to its minimum size. This will only resize the filesystem, not its containing block device, but you can resize the partition afterwards, or copy the data to a new block device of the appropriate size, if you want.

If it's acceptable for the resulting filesystem to be read-only, you could use something like SquashFS , which will create the filesystem at its target size in a single step.