Previously, file_size was calculated as num_chunks * chunk_size, which incorrectly treated the file size as the full capacity of all allocated chunks. The actual file size should be the offset of the last byte written plus 1. Changes: - Added file_size field to File structure to track actual file extent - Initialize file_size to 0 when creating new files - Update file_size in file_tree_write() based on write offset + length - Modified file_tree_read() to use file_size instead of num_chunks * chunk_size - Updated serialization/deserialization to handle file_size field This ensures that actual_bytes calculations correctly reflect the true file size, not just the allocated chunk capacity.
ToastyFS
ToastyFS is a distributed file system designed for self-hosting, so it aims to be pragmatic, understandable, and robust. You can use ToastyFS to store your files reliably over multiple machines knowing they will be automatically replicated and healed in case of hardware failure. ToastyFS works by running nodes on multiple machines. Clients using the ToastyFS C library can then send file operations to the cluster. Here's a quick example:
#include <ToastyFS.h>
int main(void)
{
ToastyString addr = TOASTY_STR("127.0.0.1");
int port = 8080;
// Connect to cluster
ToastyFS *toasty = toasty_connect(addr, port);
ToastyString file = TOASTY_STR("/my_file.txt");
// Create and write to a file
toasty_create_file(toasty, file, 4096);
toasty_write(toasty, file, 0, "Hello!", 6);
// Read it back
char buf[6];
toasty_read(toasty, file, 0, buf, 6);
// Done!
toasty_disconnect(toasty);
return 0;
}
⚠️ Note that ToastyFS is still in early development ⚠️
🎵 Now let's get toasty 🎵
Features
- Cross-platform (runs on Windows and Linux)
- Automatic Replication & Self-Healing
- Automatic content deduplication via internal content-addressing
- Configurable file chunk sizes
- Small and understandable
But ToastyFS is still in early development, so here are the missing features:
- No master replication
- No authentication or encryption
Testing
ToastyFS is tested by running an in-memory simulation of a cluster with many clients running hundreds of random operations in parallel. The test is run for long periods of times under valgrind or compiled with sanitizers.