Cargo.lock

@@ -462,6 +462,7 @@ dependencies = [
  "log",
  "md-5",
  "percent-encoding",
+ "rand",
  "roxmltree",
  "sha2",
  "tokio",

README.md

@@ -26,12 +26,18 @@ We propose the following quickstart to setup a full dev. environment as quickly
   4. Run `./script/dev-cluster.sh` to launch a test cluster (feel free to read the script)
   5. Run `./script/dev-configure.sh` to configure your test cluster with default values (same datacenter, 100 tokens)
   6. Run `./script/dev-bucket.sh` to create a bucket named `eprouvette` and an API key that will be stored in `/tmp/garage.s3`
-  7. Run `source ./script/dev-env.sh` to configure your CLI environment
+  7. Run `source ./script/dev-env-aws.sh` to configure your CLI environment
   8. You can use `garage` to manage the cluster. Try `garage --help`.
-  9. You can use `s3grg` to add, remove, and delete files. Try `s3grg --help`, `s3grg cp /proc/cpuinfo s3://eprouvette/cpuinfo.txt`, `s3grg ls s3://eprouvette`. `s3grg` is a wrapper on the `aws s3` subcommand configured with the previously generated API key (the one in `/tmp/garage.s3`).
+  9. You can use the `awsgrg` alias to add, remove, and delete files. Try `awsgrg help`, `awsgrg cp /proc/cpuinfo s3://eprouvette/cpuinfo.txt`, or `awsgrg ls s3://eprouvette`. `awsgrg` is a wrapper on the `aws s3` command pre-configured with the previously generated API key (the one in `/tmp/garage.s3`) and localhost as the endpoint.
 
 Now you should be ready to start hacking on garage!
 
+## S3 compatibility
+
+Only a subset of S3 is supported: adding, listing, getting and deleting files in a bucket.
+Bucket management, ACL and other advanced features are not (yet?) handled through the S3 API but through the `garage` CLI.
+We primarily test `garage` against the `awscli` tool and `nextcloud`.
+
 ## Setting up Garage
 
 Use the `genkeys.sh` script to generate TLS keys for encrypting communications between Garage nodes.

script/dev-env-aws.sh

@@ -0,0 +1,14 @@
+#!/bin/bash
+
+SCRIPT_FOLDER="`dirname \"${BASH_SOURCE[0]}\"`"
+REPO_FOLDER="${SCRIPT_FOLDER}/../"
+GARAGE_DEBUG="${REPO_FOLDER}/target/debug/"
+GARAGE_RELEASE="${REPO_FOLDER}/target/release/"
+PATH="${GARAGE_DEBUG}:${GARAGE_RELEASE}:$PATH"
+
+export AWS_ACCESS_KEY_ID=`cat /tmp/garage.s3 |cut -d' ' -f1`
+export AWS_SECRET_ACCESS_KEY=`cat /tmp/garage.s3 |cut -d' ' -f2`
+export AWS_DEFAULT_REGION='garage'
+
+alias awsgrg="aws s3 \
+  --endpoint-url http://127.0.0.1:3911"

script/test-smoke.sh

@@ -11,12 +11,52 @@ ${SCRIPT_FOLDER}/dev-clean.sh
 ${SCRIPT_FOLDER}/dev-cluster.sh > /tmp/garage.log 2>&1 &
 ${SCRIPT_FOLDER}/dev-configure.sh
 ${SCRIPT_FOLDER}/dev-bucket.sh
-source ${SCRIPT_FOLDER}/dev-env.sh
+source ${SCRIPT_FOLDER}/dev-env-aws.sh
+source ${SCRIPT_FOLDER}/dev-env-s3cmd.sh
 
-dd if=/dev/urandom of=/tmp/garage.rnd bs=1M count=10
+garage status
+garage key list
+garage bucket list
+
+dd if=/dev/urandom of=/tmp/garage.1.rnd bs=1k count=2 # < INLINE_THRESHOLD = 3072 bytes
+dd if=/dev/urandom of=/tmp/garage.2.rnd bs=1M count=5
+dd if=/dev/urandom of=/tmp/garage.3.rnd bs=1M count=10
+
+for idx in $(seq 1 3); do
+  # AWS sends	
+  awsgrg cp /tmp/garage.$idx.rnd s3://eprouvette/garage.$idx.aws
+  
+  awsgrg ls s3://eprouvette
+  
+  awsgrg cp s3://eprouvette/garage.$idx.aws /tmp/garage.$idx.dl
+  diff /tmp/garage.$idx.rnd /tmp/garage.$idx.dl
+  rm /tmp/garage.$idx.dl
+  
+  s3grg get s3://eprouvette/garage.$idx.aws /tmp/garage.$idx.dl
+  diff /tmp/garage.$idx.rnd /tmp/garage.$idx.dl
+  rm /tmp/garage.$idx.dl
+  
+  awsgrg rm s3://eprouvette/garage.$idx.aws
+
+  # S3CMD sends
+  s3grg put /tmp/garage.$idx.rnd s3://eprouvette/garage.$idx.s3cmd
 
-s3grg cp /tmp/garage.rnd s3://eprouvette/
   s3grg ls s3://eprouvette
-s3grg cp s3://eprouvette/garage.rnd /tmp/garage.dl
 	
-diff /tmp/garage.rnd /tmp/garage.dl
+  s3grg get s3://eprouvette/garage.$idx.s3cmd /tmp/garage.$idx.dl
+  diff /tmp/garage.$idx.rnd /tmp/garage.$idx.dl
+  rm /tmp/garage.$idx.dl
+  
+  awsgrg cp s3://eprouvette/garage.$idx.s3cmd /tmp/garage.$idx.dl
+  diff /tmp/garage.$idx.rnd /tmp/garage.$idx.dl
+  rm /tmp/garage.$idx.dl
+	
+  s3grg rm s3://eprouvette/garage.$idx.s3cmd
+done
+rm /tmp/garage.{1,2,3}.rnd
+
+garage bucket deny --read --write eprouvette --key $AWS_ACCESS_KEY_ID
+garage bucket delete --yes eprouvette
+garage key delete --yes $AWS_ACCESS_KEY_ID
+
+echo "success"

src/api/Cargo.toml

@@ -27,6 +27,7 @@ md-5 = "0.9.1"
 sha2 = "0.8"
 hmac = "0.7"
 crypto-mac = "0.7"
+rand = "0.7"
 
 futures = "0.3"
 futures-util = "0.3"

src/api/s3_get.rs

@@ -24,10 +24,13 @@ fn object_headers(
 			"Content-Type",
 			version_meta.headers.content_type.to_string(),
 		)
-		.header("ETag", version_meta.etag.to_string())
 		.header("Last-Modified", date_str)
 		.header("Accept-Ranges", format!("bytes"));
 
+	if !version_meta.etag.is_empty() {
+		resp = resp.header("ETag", format!("\"{}\"", version_meta.etag));
+	}
+
 	for (k, v) in version_meta.headers.other.iter() {
 		resp = resp.header(k, v.to_string());
 	}

src/api/s3_list.rs

@@ -18,6 +18,7 @@ use crate::encoding::*;
 struct ListResultInfo {
 	last_modified: u64,
 	size: u64,
+	etag: String,
 }
 
 pub async fn handle_list(
@@ -56,12 +57,12 @@ pub async fn handle_list(
 
 		for object in objects.iter() {
 			if !object.key.starts_with(prefix) {
-				truncated = false;
+				truncated = None;
 				break 'query_loop;
 			}
 			if let Some(version) = object.versions().iter().find(|x| x.is_data()) {
 				if result_keys.len() + result_common_prefixes.len() >= max_keys {
-					truncated = true;
+					truncated = Some(object.key.to_string());
 					break 'query_loop;
 				}
 				let common_prefix = if delimiter.len() > 0 {
@@ -75,19 +76,18 @@ pub async fn handle_list(
 				if let Some(pfx) = common_prefix {
 					result_common_prefixes.insert(pfx.to_string());
 				} else {
-					let size = match &version.state {
-						ObjectVersionState::Complete(ObjectVersionData::Inline(meta, _)) => {
-							meta.size
-						}
+					let meta = match &version.state {
+						ObjectVersionState::Complete(ObjectVersionData::Inline(meta, _)) => meta,
 						ObjectVersionState::Complete(ObjectVersionData::FirstBlock(meta, _)) => {
-							meta.size
+							meta
 						}
 						_ => unreachable!(),
 					};
 					let info = match result_keys.get(&object.key) {
 						None => ListResultInfo {
 							last_modified: version.timestamp,
-							size,
+							size: meta.size,
+							etag: meta.etag.to_string(),
 						},
 						Some(_lri) => {
 							return Err(Error::Message(format!("Duplicate key?? {}", object.key)))
@@ -98,7 +98,7 @@ pub async fn handle_list(
 			}
 		}
 		if objects.len() < max_keys + 1 {
-			truncated = false;
+			truncated = None;
 			break 'query_loop;
 		}
 		if objects.len() > 0 {
@@ -113,11 +113,22 @@ pub async fn handle_list(
 		r#"<ListBucketResult xmlns="http://s3.amazonaws.com/doc/2006-03-01/">"#
 	)
 	.unwrap();
-	writeln!(&mut xml, "\t<Bucket>{}</Bucket>", bucket).unwrap();
+	writeln!(&mut xml, "\t<Name>{}</Name>", bucket).unwrap();
 	writeln!(&mut xml, "\t<Prefix>{}</Prefix>", prefix).unwrap();
+	if let Some(mkr) = marker {
+		writeln!(&mut xml, "\t<Marker>{}</Marker>", mkr).unwrap();
+	}
 	writeln!(&mut xml, "\t<KeyCount>{}</KeyCount>", result_keys.len()).unwrap();
 	writeln!(&mut xml, "\t<MaxKeys>{}</MaxKeys>", max_keys).unwrap();
-	writeln!(&mut xml, "\t<IsTruncated>{}</IsTruncated>", truncated).unwrap();
+	writeln!(
+		&mut xml,
+		"\t<IsTruncated>{}</IsTruncated>",
+		truncated.is_some()
+	)
+	.unwrap();
+	if let Some(next_marker) = truncated {
+		writeln!(&mut xml, "\t<NextMarker>{}</NextMarker>", next_marker).unwrap();
+	}
 	for (key, info) in result_keys.iter() {
 		let last_modif = NaiveDateTime::from_timestamp(info.last_modified as i64 / 1000, 0);
 		let last_modif = DateTime::<Utc>::from_utc(last_modif, Utc);
@@ -132,6 +143,9 @@ pub async fn handle_list(
 		.unwrap();
 		writeln!(&mut xml, "\t\t<LastModified>{}</LastModified>", last_modif).unwrap();
 		writeln!(&mut xml, "\t\t<Size>{}</Size>", info.size).unwrap();
+		if !info.etag.is_empty() {
+			writeln!(&mut xml, "\t\t<ETag>\"{}\"</ETag>", info.etag).unwrap();
+		}
 		writeln!(&mut xml, "\t\t<StorageClass>STANDARD</StorageClass>").unwrap();
 		writeln!(&mut xml, "\t</Contents>").unwrap();
 	}

src/api/s3_put.rs

@@ -428,6 +428,21 @@ pub async fn handle_complete_multipart_upload(
 		_ => unreachable!(),
 	};
 
+	// ETag calculation: we produce ETags that have the same form as
+	// those of S3 multipart uploads, but we don't use their actual
+	// calculation for the first part (we use random bytes). This
+	// shouldn't impact compatibility as the S3 docs specify that
+	// the ETag is an opaque value in case of a multipart upload.
+	// See also: https://teppen.io/2018/06/23/aws_s3_etags/
+	let num_parts = version.blocks().last().unwrap().part_number
+		- version.blocks().first().unwrap().part_number
+		+ 1;
+	let etag = format!(
+		"{}-{}",
+		hex::encode(&rand::random::<[u8; 16]>()[..]),
+		num_parts
+	);
+
 	// TODO: check that all the parts that they pretend they gave us are indeed there
 	// TODO: when we read the XML from _req, remember to check the sha256 sum of the payload
 	//       against the signed x-amz-content-sha256
@@ -442,7 +457,7 @@ pub async fn handle_complete_multipart_upload(
 		ObjectVersionMeta {
 			headers,
 			size: total_size,
-			etag: "".to_string(), // TODO
+			etag: etag,
 		},
 		version.blocks()[0].hash,
 	));

src/table/table.rs

@@ -391,7 +391,8 @@ where
 			let (old_entry, new_entry) = self.store.transaction(|db| {
 				let (old_entry, new_entry) = match db.get(&tree_key)? {
 					Some(prev_bytes) => {
-						let old_entry = self.decode_entry(&prev_bytes)
+						let old_entry = self
+							.decode_entry(&prev_bytes)
 							.map_err(sled::ConflictableTransactionError::Abort)?;
 						let mut new_entry = old_entry.clone();
 						new_entry.merge(&update);