Abstract database behind generic interface and implement alternative drivers (#322)

- [x] Design interface
- [x] Implement Sled backend
  - [x] Re-implement the SledCountedTree hack ~~on Sled backend~~ on all backends (i.e. over the abstraction)
- [x] Convert Garage code to use generic interface
- [x] Proof-read converted Garage code
- [ ] Test everything well
- [x] Implement sqlite backend
- [x] Implement LMDB backend
- [ ] (Implement Persy backend?)
- [ ] (Implement other backends? (like RocksDB, ...))
- [x] Implement backend choice in config file and garage server module
- [x] Add CLI for converting between DB formats
- Exploit the new interface to put more things in transactions
  - [x] `.updated()` trigger on Garage tables

Fix #284

**Bugs**

- [x] When exporting sqlite, trees iterate empty??
- [x] LMDB doesn't work

**Known issues for various back-ends**

- Sled:
  - Eats all my RAM and also all my disk space
  - `.len()` has to traverse the whole table
  - Is actually quite slow on some operations
  - And is actually pretty bad code...
- Sqlite:
  - Requires a lock to be taken on all operations. The lock is also taken when iterating on a table with `.iter()`, and the lock isn't released until the iterator is dropped. This means that we must be VERY carefull to not do anything else inside a `.iter()` loop or else we will have a deadlock! Most such cases have been eliminated from the Garage codebase, but there might still be some that remain. If your Garage-over-Sqlite seems to hang/freeze, this is the reason.
  - (adapter uses a bunch of unsafe code)
- Heed (LMDB):
  - Not suited for 32-bit machines as it has to map the whole DB in memory.
  - (adpater uses a tiny bit of unsafe code)

**My recommendation:** avoid 32-bit machines and use LMDB as much as possible.

**Converting databases** is actually quite easy. For example from Sled to LMDB:

```bash
cd src/db
cargo run --features cli --bin convert -- -i path/to/garage/meta/db -a sled -o path/to/garage/meta/db.lmdb -b lmdb
```

Then, just add this to your `config.toml`:

```toml
db_engine = "lmdb"
```

Co-authored-by: Alex Auvolat <alex@adnab.me>
Reviewed-on: https://git.deuxfleurs.fr/Deuxfleurs/garage/pulls/322
Co-authored-by: Alex <alex@adnab.me>
Co-committed-by: Alex <alex@adnab.me>

5 files changed, 19 insertions, 110 deletions

diff --git a/src/util/Cargo.toml b/src/util/Cargo.toml
index 95cde531..5d073436 100644
--- a/src/util/Cargo.toml
+++ b/src/util/Cargo.toml
@@ -14,6 +14,8 @@ path = "lib.rs"
 # See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
 
 [dependencies]
+garage_db = { version = "0.8.0", path = "../db" }
+
 blake2 = "0.9"
 err-derive = "0.3"
 xxhash-rust = { version = "0.8", default-features = false, features = ["xxh3"] }
@@ -22,8 +24,6 @@ tracing = "0.1.30"
 rand = "0.8"
 sha2 = "0.9"
 
-sled = "0.34"
-
 chrono = "0.4"
 rmp-serde = "0.15"
 serde = { version = "1.0", default-features = false, features = ["derive", "rc"] }
diff --git a/src/util/config.rs b/src/util/config.rs
index 99ebce31..e8ef4fdd 100644
--- a/src/util/config.rs
+++ b/src/util/config.rs
@@ -64,14 +64,19 @@ pub struct Config {
 	#[serde(default)]
 	pub kubernetes_skip_crd: bool,
 
+	// -- DB
+	/// Database engine to use for metadata (options: sled, sqlite, lmdb)
+	#[serde(default = "default_db_engine")]
+	pub db_engine: String,
+
 	/// Sled cache size, in bytes
 	#[serde(default = "default_sled_cache_capacity")]
 	pub sled_cache_capacity: u64,
-
 	/// Sled flush interval in milliseconds
 	#[serde(default = "default_sled_flush_every_ms")]
 	pub sled_flush_every_ms: u64,
 
+	// -- APIs
 	/// Configuration for S3 api
 	pub s3_api: S3ApiConfig,
 
@@ -129,6 +134,10 @@ pub struct AdminConfig {
 	pub trace_sink: Option<String>,
 }
 
+fn default_db_engine() -> String {
+	"sled".into()
+}
+
 fn default_sled_cache_capacity() -> u64 {
 	128 * 1024 * 1024
 }
diff --git a/src/util/error.rs b/src/util/error.rs
index 8734a0c8..9995c746 100644
--- a/src/util/error.rs
+++ b/src/util/error.rs
@@ -26,8 +26,8 @@ pub enum Error {
 	#[error(display = "Netapp error: {}", _0)]
 	Netapp(#[error(source)] netapp::error::Error),
 
-	#[error(display = "Sled error: {}", _0)]
-	Sled(#[error(source)] sled::Error),
+	#[error(display = "DB error: {}", _0)]
+	Db(#[error(source)] garage_db::Error),
 
 	#[error(display = "Messagepack encode error: {}", _0)]
 	RmpEncode(#[error(source)] rmp_serde::encode::Error),
@@ -78,11 +78,11 @@ impl Error {
 	}
 }
 
-impl From<sled::transaction::TransactionError<Error>> for Error {
-	fn from(e: sled::transaction::TransactionError<Error>) -> Error {
+impl From<garage_db::TxError<Error>> for Error {
+	fn from(e: garage_db::TxError<Error>) -> Error {
 		match e {
-			sled::transaction::TransactionError::Abort(x) => x,
-			sled::transaction::TransactionError::Storage(x) => Error::Sled(x),
+			garage_db::TxError::Abort(x) => x,
+			garage_db::TxError::Db(x) => Error::Db(x),
 		}
 	}
 }
diff --git a/src/util/lib.rs b/src/util/lib.rs
index d8ffdd0b..8ca6e310 100644
--- a/src/util/lib.rs
+++ b/src/util/lib.rs
@@ -11,7 +11,7 @@ pub mod error;
 pub mod formater;
 pub mod metrics;
 pub mod persister;
-pub mod sled_counter;
+//pub mod sled_counter;
 pub mod time;
 pub mod token_bucket;
 pub mod tranquilizer;
diff --git a/src/util/sled_counter.rs b/src/util/sled_counter.rs
deleted file mode 100644
index bc54cea0..00000000
--- a/src/util/sled_counter.rs
+++ /dev/null
@@ -1,100 +0,0 @@
-use std::sync::{
-	atomic::{AtomicUsize, Ordering},
-	Arc,
-};
-
-use sled::{CompareAndSwapError, IVec, Iter, Result, Tree};
-
-#[derive(Clone)]
-pub struct SledCountedTree(Arc<SledCountedTreeInternal>);
-
-struct SledCountedTreeInternal {
-	tree: Tree,
-	len: AtomicUsize,
-}
-
-impl SledCountedTree {
-	pub fn new(tree: Tree) -> Self {
-		let len = tree.len();
-		Self(Arc::new(SledCountedTreeInternal {
-			tree,
-			len: AtomicUsize::new(len),
-		}))
-	}
-
-	pub fn len(&self) -> usize {
-		self.0.len.load(Ordering::Relaxed)
-	}
-
-	pub fn is_empty(&self) -> bool {
-		self.0.tree.is_empty()
-	}
-
-	pub fn get<K: AsRef<[u8]>>(&self, key: K) -> Result<Option<IVec>> {
-		self.0.tree.get(key)
-	}
-
-	pub fn iter(&self) -> Iter {
-		self.0.tree.iter()
-	}
-
-	// ---- writing functions ----
-
-	pub fn insert<K, V>(&self, key: K, value: V) -> Result<Option<IVec>>
-	where
-		K: AsRef<[u8]>,
-		V: Into<IVec>,
-	{
-		let res = self.0.tree.insert(key, value);
-		if res == Ok(None) {
-			self.0.len.fetch_add(1, Ordering::Relaxed);
-		}
-		res
-	}
-
-	pub fn remove<K: AsRef<[u8]>>(&self, key: K) -> Result<Option<IVec>> {
-		let res = self.0.tree.remove(key);
-		if matches!(res, Ok(Some(_))) {
-			self.0.len.fetch_sub(1, Ordering::Relaxed);
-		}
-		res
-	}
-
-	pub fn pop_min(&self) -> Result<Option<(IVec, IVec)>> {
-		let res = self.0.tree.pop_min();
-		if let Ok(Some(_)) = &res {
-			self.0.len.fetch_sub(1, Ordering::Relaxed);
-		};
-		res
-	}
-
-	pub fn compare_and_swap<K, OV, NV>(
-		&self,
-		key: K,
-		old: Option<OV>,
-		new: Option<NV>,
-	) -> Result<std::result::Result<(), CompareAndSwapError>>
-	where
-		K: AsRef<[u8]>,
-		OV: AsRef<[u8]>,
-		NV: Into<IVec>,
-	{
-		let old_some = old.is_some();
-		let new_some = new.is_some();
-
-		let res = self.0.tree.compare_and_swap(key, old, new);
-
-		if res == Ok(Ok(())) {
-			match (old_some, new_some) {
-				(false, true) => {
-					self.0.len.fetch_add(1, Ordering::Relaxed);
-				}
-				(true, false) => {
-					self.0.len.fetch_sub(1, Ordering::Relaxed);
-				}
-				_ => (),
-			}
-		}
-		res
-	}
-}