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>
This commit is contained in:
Alex
2022-06-08 10:01:44 +02:00
parent 7eed3ceda9
commit b44d3fc796
42 changed files with 3086 additions and 641 deletions
+29 -31
View File
@@ -6,6 +6,8 @@ use std::time::Duration;
use serde::{Deserialize, Serialize};
use tokio::sync::{mpsc, watch};
use garage_db as db;
use garage_rpc::ring::Ring;
use garage_rpc::system::System;
use garage_util::data::*;
@@ -114,10 +116,6 @@ impl<T: CounterSchema> TableSchema for CounterTable<T> {
type E = CounterEntry<T>;
type Filter = (DeletedFilter, Vec<Uuid>);
fn updated(&self, _old: Option<&Self::E>, _new: Option<&Self::E>) {
// nothing for now
}
fn matches_filter(entry: &Self::E, filter: &Self::Filter) -> bool {
if filter.0 == DeletedFilter::Any {
return true;
@@ -135,7 +133,7 @@ impl<T: CounterSchema> TableSchema for CounterTable<T> {
pub struct IndexCounter<T: CounterSchema> {
this_node: Uuid,
local_counter: sled::Tree,
local_counter: db::Tree,
propagate_tx: mpsc::UnboundedSender<(T::P, T::S, LocalCounterEntry)>,
pub table: Arc<Table<CounterTable<T>, TableShardedReplication>>,
}
@@ -144,7 +142,7 @@ impl<T: CounterSchema> IndexCounter<T> {
pub fn new(
system: Arc<System>,
replication: TableShardedReplication,
db: &sled::Db,
db: &db::Db,
) -> Arc<Self> {
let background = system.background.clone();
@@ -174,36 +172,36 @@ impl<T: CounterSchema> IndexCounter<T> {
this
}
pub fn count(&self, pk: &T::P, sk: &T::S, counts: &[(&str, i64)]) -> Result<(), Error> {
pub fn count(
&self,
tx: &mut db::Transaction,
pk: &T::P,
sk: &T::S,
counts: &[(&str, i64)],
) -> db::TxResult<(), Error> {
let tree_key = self.table.data.tree_key(pk, sk);
let new_entry = self.local_counter.transaction(|tx| {
let mut entry = match tx.get(&tree_key[..])? {
Some(old_bytes) => {
rmp_serde::decode::from_read_ref::<_, LocalCounterEntry>(&old_bytes)
.map_err(Error::RmpDecode)
.map_err(sled::transaction::ConflictableTransactionError::Abort)?
}
None => LocalCounterEntry {
values: BTreeMap::new(),
},
};
let mut entry = match tx.get(&self.local_counter, &tree_key[..])? {
Some(old_bytes) => rmp_serde::decode::from_read_ref::<_, LocalCounterEntry>(&old_bytes)
.map_err(Error::RmpDecode)
.map_err(db::TxError::Abort)?,
None => LocalCounterEntry {
values: BTreeMap::new(),
},
};
for (s, inc) in counts.iter() {
let mut ent = entry.values.entry(s.to_string()).or_insert((0, 0));
ent.0 += 1;
ent.1 += *inc;
}
for (s, inc) in counts.iter() {
let mut ent = entry.values.entry(s.to_string()).or_insert((0, 0));
ent.0 += 1;
ent.1 += *inc;
}
let new_entry_bytes = rmp_to_vec_all_named(&entry)
.map_err(Error::RmpEncode)
.map_err(sled::transaction::ConflictableTransactionError::Abort)?;
tx.insert(&tree_key[..], new_entry_bytes)?;
let new_entry_bytes = rmp_to_vec_all_named(&entry)
.map_err(Error::RmpEncode)
.map_err(db::TxError::Abort)?;
tx.insert(&self.local_counter, &tree_key[..], new_entry_bytes)?;
Ok(entry)
})?;
if let Err(e) = self.propagate_tx.send((pk.clone(), sk.clone(), new_entry)) {
if let Err(e) = self.propagate_tx.send((pk.clone(), sk.clone(), entry)) {
error!(
"Could not propagate updated counter values, failed to send to channel: {}",
e