Fix tests

Signed-off-by: Heinz N. Gies <[email protected]>

src/connectors/prelude.rs

@@ -46,7 +46,7 @@ pub(crate) use tremor_common::{
 };
 pub(crate) use tremor_config::Impl;
 pub use tremor_config::NameWithConfig;
-pub use tremor_pipeline::{CbAction,  EventOriginUri, DEFAULT_STREAM_ID};
+pub use tremor_pipeline::{CbAction, EventOriginUri, DEFAULT_STREAM_ID};
 pub(crate) use tremor_script::prelude::*;
 /// default buf size used for reading from files and streams (sockets etc)
 ///

src/raft/store.rs

@@ -477,6 +477,13 @@ impl Store {
     where
         T: for<'de> serde::Deserialize<'de>,
     {
+        // We need to use a write transaction despite just wanting a read transaction due to
+        // https://github.com/cberner/redb/issues/711
+        let bug_fix_txn = self.db.begin_write().map_err(w_err)?;
+        {
+            let _ = bug_fix_txn.open_table(STORE).map_err(w_err)?;
+        }
+        bug_fix_txn.commit().map_err(w_err)?;
         let read_txn = self.db.begin_read().map_err(r_err)?;
         let table = read_txn.open_table(STORE).map_err(r_err)?;
         let r = table.get(key).map_err(r_err)?;
@@ -528,6 +535,13 @@ impl RaftLogReader<TremorRaftConfig> for Store {
         &mut self,
         range: RB,
     ) -> StorageResult<Vec<Entry<TremorRaftConfig>>> {
+        // We need to use a write transaction despite just wanting a read transaction due to
+        // https://github.com/cberner/redb/issues/711
+        let bug_fix_txn = self.db.begin_write().map_err(w_err)?;
+        {
+            let _ = bug_fix_txn.open_table(LOGS).map_err(w_err)?;
+        }
+        bug_fix_txn.commit().map_err(w_err)?;
         let read_txn = self.db.begin_read().map_err(r_err)?;
 
         let table = read_txn.open_table(LOGS).map_err(r_err)?;
@@ -770,6 +784,14 @@ impl RaftStorage<TremorRaftConfig> for Store {
     }
 
     async fn get_log_state(&mut self) -> StorageResult<LogState<TremorRaftConfig>> {
+        // We need to use a write transaction despite just wanting a read transaction due to
+        // https://github.com/cberner/redb/issues/711
+        let bug_fix_txn = self.db.begin_write().map_err(w_err)?;
+        {
+            let _ = bug_fix_txn.open_table(LOGS).map_err(w_err)?;
+        }
+        bug_fix_txn.commit().map_err(w_err)?;
+
         let read_txn = self.db.begin_read().map_err(r_err)?;
         let table = read_txn.open_table(LOGS).map_err(r_err)?;
         let last = table
@@ -894,6 +916,14 @@ impl Store {
     /// # Errors
     /// if the store fails to read the RPC address
     pub fn get_self_addr(db: &Database) -> Result<Option<Addr>, Error> {
+        // We need to use a write transaction despite just wanting a read transaction due to
+        // https://github.com/cberner/redb/issues/711
+        let bug_fix_txn = db.begin_write()?;
+        {
+            let _ = bug_fix_txn.open_table(SYSTEM).map_err(w_err)?;
+        }
+        bug_fix_txn.commit().map_err(w_err)?;
+
         let read_txn = db.begin_read().map_err(r_err)?;
 
         let table = read_txn.open_table(SYSTEM).map_err(r_err)?;
@@ -904,6 +934,14 @@ impl Store {
     /// # Errors
     /// if the store fails to read the node id
     pub fn get_self_node_id(db: &Database) -> Result<Option<crate::raft::NodeId>, Error> {
+        // We need to use a write transaction despite just wanting a read transaction due to
+        // https://github.com/cberner/redb/issues/711
+        let bug_fix_txn = db.begin_write()?;
+        {
+            let _ = bug_fix_txn.open_table(SYSTEM).map_err(w_err)?;
+        }
+        bug_fix_txn.commit().map_err(w_err)?;
+
         let read_txn = db.begin_read().map_err(r_err)?;
 
         let table = read_txn.open_table(SYSTEM).map_err(r_err)?;

src/raft/store/statemachine.rs

@@ -164,6 +164,14 @@ impl TremorStateMachine {
     where
         T: for<'de> serde::Deserialize<'de>,
     {
+        // We need to use a write transaction despite just wanting a read transaction due to
+        // https://github.com/cberner/redb/issues/711
+        let bug_fix_txn = self.db.begin_write().map_err(w_err)?;
+        {
+            let _ = bug_fix_txn.open_table(STATE_MACHINE).map_err(w_err)?;
+        }
+        bug_fix_txn.commit().map_err(w_err)?;
+
         let read_txn = self.db.begin_read().map_err(r_err)?;
 
         let table = read_txn.open_table(STATE_MACHINE).map_err(r_err)?;

src/raft/store/statemachine/apps.rs

@@ -90,6 +90,14 @@ impl RaftStateMachine<AppsSnapshot, AppsRequest> for AppsStateMachine {
             apps: HashMap::new(),
             world: world.clone(),
         };
+        // We need to use a write transaction despite just wanting a read transaction due to
+        // https://github.com/cberner/redb/issues/711
+        let bug_fix_txn = db.begin_write()?;
+        {
+            let _ = bug_fix_txn.open_table(APPS).map_err(w_err)?;
+            let _ = bug_fix_txn.open_table(INSTANCES).map_err(w_err)?;
+        }
+        bug_fix_txn.commit().map_err(w_err)?;
 
         let read_txn = db.begin_read()?;
         let apps = read_txn.open_table(APPS)?;
@@ -99,6 +107,7 @@ impl RaftStateMachine<AppsSnapshot, AppsRequest> for AppsStateMachine {
             me.load_archive(archive.value())
                 .map_err(|e| store::Error::Other(Box::new(e)))?;
         }
+
         let instances = read_txn.open_table(INSTANCES)?;
 
         // load instances
@@ -220,6 +229,13 @@ impl RaftStateMachine<AppsSnapshot, AppsRequest> for AppsStateMachine {
     }
 
     fn as_snapshot(&self) -> StorageResult<AppsSnapshot> {
+        // We need to use a write transaction despite just wanting a read transaction due to
+        // https://github.com/cberner/redb/issues/711
+        let bug_fix_txn = self.db.begin_write().map_err(w_err)?;
+        {
+            let _ = bug_fix_txn.open_table(APPS).map_err(w_err)?;
+        }
+        bug_fix_txn.commit().map_err(w_err)?;
         let read_txn = self.db.begin_read().map_err(r_err)?;
         let apps = read_txn.open_table(APPS).map_err(r_err)?;
         let archives = apps
@@ -300,7 +316,7 @@ impl AppsStateMachine {
         }
         write_txn.commit().map_err(w_err)?;
 
-        let app = StateApp {
+        let app: StateApp = StateApp {
             app,
             main,
             arena_indices,

src/raft/store/statemachine/kv.rs

@@ -52,6 +52,13 @@ impl KvStateMachine {
     /// try to obtain the value at the given `key`.
     /// Returns `Ok(None)` if there is no value for that key.
     pub(crate) fn get(&self, key: &str) -> StorageResult<Option<Vec<u8>>> {
+        // We need to use a write transaction despite just wanting a read transaction due to
+        // https://github.com/cberner/redb/issues/711
+        let bug_fix_txn = self.db.begin_write().map_err(w_err)?;
+        {
+            let _ = bug_fix_txn.open_table(DATA).map_err(w_err)?;
+        }
+        bug_fix_txn.commit().map_err(w_err)?;
         let read_txn = self.db.begin_read().map_err(r_err)?;
         let table = read_txn.open_table(DATA).map_err(r_err)?;
         table
@@ -85,6 +92,14 @@ impl RaftStateMachine<KvSnapshot, KvRequest> for KvStateMachine {
     }
 
     fn as_snapshot(&self) -> StorageResult<KvSnapshot> {
+        // We need to use a write transaction despite just wanting a read transaction due to
+        // https://github.com/cberner/redb/issues/711
+        let bug_fix_txn = self.db.begin_write().map_err(w_err)?;
+        {
+            let _ = bug_fix_txn.open_table(DATA).map_err(w_err)?;
+        }
+        bug_fix_txn.commit().map_err(w_err)?;
+
         let read_tnx = self.db.begin_read().map_err(w_err)?;
         let table = read_tnx.open_table(DATA).map_err(w_err)?;
         let data = table

src/raft/store/statemachine/nodes.rs

@@ -56,6 +56,15 @@ impl RaftStateMachine<NodesSnapshot, NodesRequest> for NodesStateMachine {
         Self: std::marker::Sized,
     {
         // load known nodes
+
+        // We need to use a write transaction despite just wanting a read transaction due to
+        // https://github.com/cberner/redb/issues/711
+        let bug_fix_txn = db.begin_write().map_err(w_err)?;
+        {
+            let _ = bug_fix_txn.open_table(SYSTEM).map_err(w_err)?;
+            let _ = bug_fix_txn.open_table(NODES).map_err(w_err)?;
+        }
+        bug_fix_txn.commit().map_err(w_err)?;
         let read_txn = db.begin_read().map_err(r_err)?;
 
         let table = read_txn.open_table(SYSTEM).map_err(r_err)?;
@@ -66,6 +75,7 @@ impl RaftStateMachine<NodesSnapshot, NodesRequest> for NodesStateMachine {
                 debug!("No next_node_id stored in db, starting from 0");
                 0
             };
+
         let table = read_txn.open_table(NODES).map_err(r_err)?;
 
         let known_nodes = table

src/raft/test.rs

@@ -115,9 +115,9 @@ async fn cluster_join_test() -> ClusterResult<()> {
     let dir0 = tempfile::tempdir()?;
     let dir1 = tempfile::tempdir()?;
     let dir2 = tempfile::tempdir()?;
-    let node0 = TestNode::bootstrap(dir0.path()).await?;
-    let node1 = TestNode::start_and_join(dir1.path(), &node0.addr).await?;
-    let node2 = TestNode::start_and_join(dir2.path(), &node1.addr).await?;
+    let node0 = TestNode::bootstrap(dir0.path().join("db")).await?;
+    let node1 = TestNode::start_and_join(dir1.path().join("db"), &node0.addr).await?;
+    let node2 = TestNode::start_and_join(dir2.path().join("db"), &node1.addr).await?;
 
     // all see the same leader
     let client0 = node0.client();
@@ -162,9 +162,9 @@ async fn kill_and_restart_voter() -> ClusterResult<()> {
     let dir1 = tempfile::tempdir()?;
     let dir2 = tempfile::tempdir()?;
 
-    let node0 = TestNode::bootstrap(dir0.path()).await?;
-    let node1 = TestNode::start_and_join(dir1.path(), &node0.addr).await?;
-    let node2 = TestNode::start_and_join(dir2.path(), &node1.addr).await?;
+    let node0 = TestNode::bootstrap(dir0.path().join("db")).await?;
+    let node1 = TestNode::start_and_join(dir1.path().join("db"), &node0.addr).await?;
+    let node2 = TestNode::start_and_join(dir2.path().join("db"), &node1.addr).await?;
 
     let client0 = node0.client();
     let metrics = client0.metrics().await?;
@@ -181,7 +181,7 @@ async fn kill_and_restart_voter() -> ClusterResult<()> {
     tokio::time::sleep(Duration::from_millis(500)).await;
 
     // restart the node
-    let node1 = TestNode::just_start(dir1.path()).await?;
+    let node1 = TestNode::just_start(dir1.path().join("db")).await?;
 
     // check that the leader is available
     // TODO: solidify to guard against timing issues

src/raft/test/learner.rs

@@ -25,9 +25,9 @@ async fn add_learner_test() -> ClusterResult<()> {
     let dir1 = tempfile::tempdir()?;
     let dir2 = tempfile::tempdir()?;
     let dir3 = tempfile::tempdir()?;
-    let node0 = TestNode::bootstrap(dir0.path()).await?;
-    let node1 = TestNode::start_and_join(dir1.path(), &node0.addr).await?;
-    let node2 = TestNode::start_and_join(dir2.path(), &node1.addr).await?;
+    let node0 = TestNode::bootstrap(dir0.path().join("db")).await?;
+    let node1 = TestNode::start_and_join(dir1.path().join("db"), &node0.addr).await?;
+    let node2 = TestNode::start_and_join(dir2.path().join("db"), &node1.addr).await?;
     let client0 = node0.client();
     let metrics = client0.metrics().await?;
     let members = metrics
@@ -38,7 +38,7 @@ async fn add_learner_test() -> ClusterResult<()> {
         .expect("No nodes in membership config");
     assert_eq!(3, members.len());
 
-    let learner_node = TestNode::join_as_learner(dir3.path(), &node0.addr).await?;
+    let learner_node = TestNode::join_as_learner(dir3.path().join("db"), &node0.addr).await?;
     let (learner_node_id, learner_addr) = learner_node.running.node_data();
     // learner is known to the cluster
     let nodemap = client0.get_nodes().await?;
@@ -82,9 +82,9 @@ async fn learner_runs_app() -> ClusterResult<()> {
     let dir1 = tempfile::tempdir()?;
     let dir2 = tempfile::tempdir()?;
     let dir3 = tempfile::tempdir()?;
-    let node0 = TestNode::bootstrap(dir0.path()).await?;
-    let node1 = TestNode::start_and_join(dir1.path(), &node0.addr).await?;
-    let node2 = TestNode::start_and_join(dir2.path(), &node1.addr).await?;
+    let node0 = TestNode::bootstrap(dir0.path().join("db")).await?;
+    let node1 = TestNode::start_and_join(dir1.path().join("db"), &node0.addr).await?;
+    let node2 = TestNode::start_and_join(dir2.path().join("db"), &node1.addr).await?;
     let client0 = node0.client();
     let metrics = client0.metrics().await?;
     let members = metrics
@@ -95,7 +95,7 @@ async fn learner_runs_app() -> ClusterResult<()> {
         .expect("No nodes in membership config");
     assert_eq!(3, members.len());
 
-    let learner_node = TestNode::join_as_learner(dir3.path(), &node0.addr).await?;
+    let learner_node = TestNode::join_as_learner(dir3.path().join("db"), &node0.addr).await?;
     let (_learner_node_id, _learner_addr) = learner_node.running.node_data();
     let tmpfile = tempfile::NamedTempFile::new()?;
     let out_path = tmpfile.into_temp_path();

tremor-value/src/known_key.rs

@@ -17,7 +17,7 @@ use beef::Cow;
 use halfbrown::RawEntryMut;
 use simd_json::ObjectHasher;
 use std::fmt;
-use std::hash::{BuildHasher};
+use std::hash::BuildHasher;
 use value_trait::prelude::*;
 
 /// Well known key that can be looked up in a `Value` faster.