PR to track down CI failures (#501)

* Use an extended timeout for DKGs specifically

* Add a log statement when message-queue connection fails

* Add a 60 second keep-alive to connections

* Use zalloc for processor/message-queue/coordinator

An additional layer which protects us against edge cases with Zeroizing
(objects which don't support it or don't miss it).

* Add further logs to message-queue

* Further increase re-attempt timeouts in CI

* Remove misplaced continue inmessage-queue client

Fixes observed CI failures.

* Revert "Further increase re-attempt timeouts in CI"

This reverts commit 3723530.

common/request/src/lib.rs

@@ -49,11 +49,14 @@ pub struct Client {
 
 impl Client {
   fn connector() -> Connector {
+    let mut res = HttpConnector::new();
+    res.set_keepalive(Some(core::time::Duration::from_secs(60)));
     #[cfg(feature = "tls")]
-    let res =
-      HttpsConnectorBuilder::new().with_native_roots().https_or_http().enable_http1().build();
-    #[cfg(not(feature = "tls"))]
-    let res = HttpConnector::new();
+    let res = HttpsConnectorBuilder::new()
+      .with_native_roots()
+      .https_or_http()
+      .enable_http1()
+      .wrap_connector(res);
     res
   }
 

coordinator/Cargo.toml

@@ -32,6 +32,7 @@ frost-schnorrkel = { path = "../crypto/schnorrkel" }
 
 scale = { package = "parity-scale-codec", version = "3", default-features = false, features = ["std", "derive"] }
 
+zalloc = { path = "../common/zalloc" }
 serai-db = { path = "../common/db" }
 serai-env = { path = "../common/env" }
 

coordinator/src/main.rs

@@ -63,6 +63,10 @@ use cosign_evaluator::CosignEvaluator;
 #[cfg(test)]
 pub mod tests;
 
+#[global_allocator]
+static ALLOCATOR: zalloc::ZeroizingAlloc<std::alloc::System> =
+  zalloc::ZeroizingAlloc(std::alloc::System);
+
 #[derive(Clone)]
 pub struct ActiveTributary<D: Db, P: P2p> {
   pub spec: TributarySpec,

coordinator/src/tributary/db.rs

@@ -141,12 +141,16 @@ impl ReattemptDb {
     // 5 minutes for attempts 0 ..= 2, 10 minutes for attempts 3 ..= 5, 15 minutes for attempts > 5
     // Assumes no event will take longer than 15 minutes, yet grows the time in case there are
     // network bandwidth issues
-    let reattempt_delay = BASE_REATTEMPT_DELAY *
+    let mut reattempt_delay = BASE_REATTEMPT_DELAY *
       ((AttemptDb::attempt(txn, genesis, topic)
         .expect("scheduling re-attempt for unknown topic") /
         3) +
         1)
       .min(3);
+    // Allow more time for DKGs since they have an extra round and much more data
+    if matches!(topic, Topic::Dkg) {
+      reattempt_delay *= 4;
+    }
     let upon_block = current_block_number + reattempt_delay;
 
     let mut reattempts = Self::get(txn, genesis, upon_block).unwrap_or(vec![]);

message-queue/Cargo.toml

@@ -40,6 +40,7 @@ env_logger = { version = "0.10", default-features = false, features = ["humantim
 # Uses a single threaded runtime since this shouldn't ever be CPU-bound
 tokio = { version = "1", default-features = false, features = ["rt", "time", "io-util", "net", "macros"] }
 
+zalloc = { path = "../common/zalloc" }
 serai-db = { path = "../common/db", optional = true }
 
 serai-env = { path = "../common/env" }

message-queue/src/client.rs

@@ -68,9 +68,13 @@ impl MessageQueue {
   async fn send(socket: &mut TcpStream, msg: MessageQueueRequest) -> bool {
     let msg = borsh::to_vec(&msg).unwrap();
     let Ok(()) = socket.write_all(&u32::try_from(msg.len()).unwrap().to_le_bytes()).await else {
+      log::warn!("couldn't send the message len");
+      return false;
+    };
+    let Ok(()) = socket.write_all(&msg).await else {
+      log::warn!("couldn't write the message");
       return false;
     };
-    let Ok(()) = socket.write_all(&msg).await else { return false };
     true
   }
 
@@ -118,20 +122,32 @@ impl MessageQueue {
     'outer: loop {
       if !first {
         tokio::time::sleep(core::time::Duration::from_secs(5)).await;
-        continue;
       }
       first = false;
 
-      let Ok(mut socket) = TcpStream::connect(&self.url).await else { continue };
+      log::trace!("opening socket to message-queue for next");
+      let mut socket = match TcpStream::connect(&self.url).await {
+        Ok(socket) => socket,
+        Err(e) => {
+          log::warn!("couldn't connect to message-queue server: {e:?}");
+          continue;
+        }
+      };
+      log::trace!("opened socket for next");
 
       loop {
         if !Self::send(&mut socket, msg.clone()).await {
           continue 'outer;
         }
-        let Ok(status) = socket.read_u8().await else {
-          continue 'outer;
+        let status = match socket.read_u8().await {
+          Ok(status) => status,
+          Err(e) => {
+            log::warn!("couldn't read status u8: {e:?}");
+            continue 'outer;
+          }
         };
         // If there wasn't a message, check again in 1s
+        // TODO: Use a notification system here
         if status == 0 {
           tokio::time::sleep(core::time::Duration::from_secs(1)).await;
           continue;
@@ -143,12 +159,17 @@ impl MessageQueue {
       // Timeout after 5 seconds in case there's an issue with the length handling
       let Ok(msg) = tokio::time::timeout(core::time::Duration::from_secs(5), async {
         // Read the message length
-        let Ok(len) = socket.read_u32_le().await else {
-          return vec![];
+        let len = match socket.read_u32_le().await {
+          Ok(len) => len,
+          Err(e) => {
+            log::warn!("couldn't read len: {e:?}");
+            return vec![];
+          }
         };
         let mut buf = vec![0; usize::try_from(len).unwrap()];
         // Read the message
         let Ok(_) = socket.read_exact(&mut buf).await else {
+          log::warn!("couldn't read the message");
           return vec![];
         };
         buf

message-queue/src/main.rs

@@ -1,6 +1,3 @@
-mod messages;
-mod queue;
-
 pub(crate) use std::{
   sync::{Arc, RwLock},
   collections::HashMap,
@@ -38,6 +35,13 @@ mod clippy {
 }
 pub(crate) use self::clippy::*;
 
+mod messages;
+mod queue;
+
+#[global_allocator]
+static ALLOCATOR: zalloc::ZeroizingAlloc<std::alloc::System> =
+  zalloc::ZeroizingAlloc(std::alloc::System);
+
 // queue RPC method
 /*
   Queues a message to be delivered from a processor to a coordinator, or vice versa.

processor/Cargo.toml

@@ -54,6 +54,7 @@ log = { version = "0.4", default-features = false, features = ["std"] }
 env_logger = { version = "0.10", default-features = false, features = ["humantime"], optional = true }
 tokio = { version = "1", default-features = false, features = ["rt-multi-thread", "sync", "time", "macros"] }
 
+zalloc = { path = "../common/zalloc" }
 serai-db = { path = "../common/db", optional = true }
 serai-env = { path = "../common/env", optional = true }
 # TODO: Replace with direct usage of primitives

processor/src/main.rs

@@ -61,6 +61,10 @@ use multisigs::{MultisigEvent, MultisigManager};
 #[cfg(test)]
 mod tests;
 
+#[global_allocator]
+static ALLOCATOR: zalloc::ZeroizingAlloc<std::alloc::System> =
+  zalloc::ZeroizingAlloc(std::alloc::System);
+
 // Items which are mutably borrowed by Tributary.
 // Any exceptions to this have to be carefully monitored in order to ensure consistency isn't
 // violated.
@@ -559,6 +563,8 @@ async fn run<N: Network, D: Db, Co: Coordinator>(mut raw_db: D, network: N, mut
   loop {
     let mut txn = raw_db.txn();
 
+    log::trace!("new db txn in run");
+
     let mut outer_msg = None;
 
     tokio::select! {