Open-Source Distributed File System (OsDFS)

Overview

This Distributed File System (OsDFS) is a robust and resilient Distributed File System (DFS) designed to ensure redundancy and reliability in storing critical data. The system is implemented using MPI, where:

• Rank 0 acts as the central metadata server.
• Ranks 1 to N-1 serve as storage nodes.

The system supports file uploads, downloads, distributed search, load balancing, and failover and recovery mechanisms.

---

Features

1. File Partitioning & Three-Way Replication

• Files are split into fixed-size chunks (32 bytes each).
• Each chunk is stored on three different nodes for redundancy.
• No requirement to maintain replication in case of failures.

2. Load Balancing

• Chunks are evenly distributed across nodes.
• No node stores more than one replica of the same chunk.

3. Heartbeat Mechanism & Failure Detection

• Each node periodically sends a heartbeat signal to the metadata server.
• If a node fails to send a heartbeat for more than 3 seconds, it is marked as down.

4. Failover & Recovery Operations

• Failover: Simulates the failure of a storage node, stopping requests and heartbeats.
• Recovery: Restores a failed node and its ability to process requests and send heartbeats.

5. File Operations

• Upload: Files are partitioned and stored with replication.
• Download: Chunks are retrieved and reassembled.
• Search: Locates file chunks containing a specific word.

6. Error Handling

• If an error occurs during operations, output -1.
• Examples: Searching for a non-existent file, downloading a file with missing chunks, etc.

---

Commands & Input Format

Commands are taken from stdin:

• upload <file_name> <absolute_file_path> – Uploads a file to ADFS.
• retrieve <file_name> – Downloads a file and reassembles it.
• search <file_name> <word> – Searches for a word in a file.
• list file <file_name> – Lists nodes storing chunks of a file.
• failover <rank> – Simulates failure of a node.
• recover <rank> – Restores a previously failed node.
• exit – Terminates the program.

---

Output Format

• Success outputs 1.
• Errors output -1.
• Specific outputs for commands such as list file, search, and retrieve.

---

Additional Notes

• The metadata server (Rank 0) is always operational.
• Maximum file size: ~1-2MB.
• Heartbeat messages sent every 1-2 seconds.
• Search functionality provides exact word matches only.

---

Example Usage

Test Case 1 (File Upload & Retrieval)

Input:

upload a.txt testcases/a.txt
list file a.txt
failover 1
failover 2
retrieve a.txt
search a.txt abcd
exit

Output:

1
0 3 1 2 3
1 3 1 2 4
2 3 1 3 4
0 3 1 2 3
1 3 1 2 4
2 3 1 3 4
1
1
abcd efgh ijkl abcd abcd defg dsds abcddufhef ebhfeihf udefheifh abcde
3
0 15 20

Test Case 2 (Failover & Recovery)

Input:

upload a.txt testcases/a.txt
failover 1
list file a.txt
recover 1
list file a.txt
exit

Output:

1
0 3 1 2 3
1 3 1 2 4
2 3 1 3 4
1
0 2 2 3
1 2 2 4
2 2 3 4
1
0 3 1 2 3
1 3 1 2 4
2 3 1 3 4

---

Implementation Details

1. Metadata Server (Iron Man - Rank 0)

• Maintains a mapping of file chunks to storage nodes.
• Handles requests for uploads, downloads, and searches.
• Monitors the health of storage nodes through heartbeats.

2. Storage Nodes (Avengers - Rank 1 to N-1)

• Store file chunks assigned by the metadata server.
• Respond to retrieval and search requests.
• Send periodic heartbeat messages to the metadata server.

3. Failover & Recovery Handling

• Nodes marked as failed do not respond to any requests.
• Recovery is simulated by restoring a node’s ability to process requests and send heartbeats.