About

I have been attempting to pass the Maelstrom distributed system challenges from fly.io. These tests use Maelstrom, a "workbench for learning distributed systems by writing your own", the Github of which can be found here. I wrote the code in Rust, using Serde & serde_json. Other than that, only the Rust standard library was used. In order to test a binary with Maelstrom, follow the directions for installing Maelstrom found in either link above. Once a binary is built, maelstrom is invoked with the path to the binary, as well as additional args to configure the test.

Challenge #1: Echo

---

For challenge 1, the node_runtime fn should look like this:

node_runtime::<echo::EchoBody, echo::EchoNode>(node, tx, rx);

Challenge directions can be found here. The following test given in the fly.io directions passes:

./maelstrom test -w echo --bin /mnt/c/Users/dstern/Documents/Dev/Practice_Code/Github_Projects/event-horizon/target/debug/event-horizon --node-count 1 --time-limit 10

Challenge #2: Unique ID Generation

---

For challenge 2, the node_runtime fn should look like this:

node_runtime::<generate_id::GenerateGuidBody, generate_id::GenerateGuidNode>(node, tx, rx);

Challenge directions can be found here. The following test given in the fly.io directions passes:

 ./maelstrom test -w unique-ids --bin /mnt/c/Users/dstern/Documents/Dev/Practice_Code/Github_Projects/event-horizon/target/debug/event-horizon --time-limit 30 --rate 1000 --node-count 3 --availability total --nemesis partition

Challenge 3a: Single-Node Broadcast

---

For challenge 3, the node_runtime fn should look like this:

node_runtime::<broadcast::BroadcastBody, broadcast::BroadcastNode>(node, tx, rx);

Challenge directions can be found here. The following test given in the fly.io directions passes:

./maelstrom test -w broadcast --bin /mnt/c/Users/dstern/Documents/Dev/Practice_Code/Github_Projects/event-horizon/target/debug/event-horizon --node-count 1 --time-limit 20 --rate 10

Challenge 3b: Multi-Node Broadcast

Challenge directions can be found here. The following test given in the fly.io directions passes:

./maelstrom test -w broadcast --bin /mnt/c/Users/dstern/Documents/Dev/Practice_Code/Github_Projects/event-horizon/target/debug/event-horizon --node-count 5 --time-limit 20 --rate 10

Challenge 3c: Fault Tolerant Broadcast

Challenge directions can be found here. The following test given in the fly.io directions passes:

./maelstrom test -w broadcast --bin /mnt/c/Users/dstern/Documents/Dev/Practice_Code/Github_Projects/event-horizon/target/debug/event-horizon --node-count 5 --time-limit 20 --rate 10 --nemesis partition

Challenge 4: Grow-Only Counter

---

For challenge 4, the node_runtime fn should look like this:

node_runtime::<grow_counter::CounterBody, grow_counter::CounterNode>(node, tx, rx);

Challenge directions can be found here. While the directions instruct the programmer to use a "sequentially-consistent key/value store service provided by Maelstrom", I chose to simply send updates of each node's counter values every second to the other nodes. The following test given in the fly.io directions passes:

./maelstrom test -w g-counter --bin /mnt/c/Users/dstern/Documents/Dev/Practice_Code/Github_Projects/event-horizon/target/debug/event-horizon --node-count 3 --rate 100 --time-limit 20 --nemesis partition

Challenge 5a: Single-Node Kafka-Style Log

---

For challenge 5, the node_runtime fn should look like this:

node_runtime::<kafka::KafkaBody, kafka::KafkaNode>(node, tx, rx);

Challenge directions can be found here. The following test given in the fly.io directions passes:

./maelstrom test -w kafka --bin /mnt/c/Users/dstern/Documents/Dev/Practice_Code/Github_Projects/event-horizon/target/debug/event-horizon --node-count 1 --concurrency 2n --time-limit 20 --rate 1000

Challenge 6a: Single-Node, Totally Available Transactions

---

For challenge 6, the node_runtime fn should look like this:

node_runtime::<kv_store::KVStoreBody, kv_store::KVStoreNode>(node, tx, rx);

Challenge directions can be found here. The following test given in the fly.io directions passes:

./maelstrom test -w txn-rw-register --bin /mnt/c/Users/dstern/Documents/Dev/Practice_Code/Github_Projects/event-horizon/target/debug/event-horizon --node-count 1 --time-limit 20 --rate 1000 --concurrency 2n --consistency-models read-uncommitted --availability total

Challenge 6b: Totally-Available, Read Uncommitted Transactions

---

Challenge directions can be found here. The following two tests given in the fly.io directions passes:

Test1:

./maelstrom test -w txn-rw-register --bin /mnt/c/Users/dstern/Documents/Dev/Practice_Code/Github_Projects/event-horizon/target/debug/event-horizon --node-count 2 --concurrency 2n --time-limit 20 --rate 1000 --consistency-models read-uncommitted

Test2:

./maelstrom test -w txn-rw-register --bin /mnt/c/Users/dstern/Documents/Dev/Practice_Code/Github_Projects/event-horizon/target/debug/event-horizon --node-count 2 --concurrency 2n --time-limit 20 --rate 1000 --consistency-models read-uncommitted --availability total --nemesis partition

Both of the above tests pass, but a note in the fly.io directions says that "There’s currently an issue in the Maelstrom checker that prohibits detection of G0 anomalies." Hopefully when this bug is fixed I can run this set of tests again.

Challenge 6C: Totally-Available, Read Committed Transactions

---

Challenge directions can be found here. The following test given in the fly.io directions passes:

./maelstrom test -w txn-rw-register --bin /mnt/c/Users/dstern/Documents/Dev/Practice_Code/Github_Projects/event-horizon/target/debug/event-horizon --node-count 2 --concurrency 2n --time-limit 20 --rate 1000 --consistency-models read-committed --availability total –-nemesis partition