Coordinator: join the cluster when NATS comes up, report the connection

A node that started while NATS was down never got a coordinator. The
connect failed at boot, bootstrap recorded a build error and the node ran
with Coordinator::None until restarted. It had no broadcast subscriber
or publisher, so cross-node push and cache invalidation to it stayed
broken, and its healthcheck said nothing about it. Losing NATS after
startup was silent too.

- The NATS client now connects in the background
  (retry_on_initial_connect): startup never waits on NATS or fails over
  it, the node gets its coordinator, subscriber and publisher at once,
  and the client keeps trying (async-nats's backoff, at most 4 s apart)
  until NATS answers. Subscriptions made meanwhile start delivering when
  it does. A configured maxReconnects still ends the attempts.
- Three new events report the connection: cluster.coordinator-connected
  (info), cluster.coordinator-disconnected (warn: lost, closed, gave up,
  or not connected within the connection timeout at startup) and
  cluster.coordinator-error (warn: a failed attempt, reported once per
  outage rather than every retry, and server errors, slow consumers and
  lame duck mode). They are in the packaged schema, ids 644 to 646.
- GET /healthz/cluster reports the coordinator: 200
  {"coordinator":"connected"}, 503 {"coordinator":"disconnected"}, or
  200 with "none" (no coordinator) or "unknown" (a backend that doesn't
  track its connection). /healthz/live and /healthz/ready are unchanged
  on purpose: a node without its coordinator still serves mail, and
  failing those would have orchestrators restart, or pull out of
  service, every node at once whenever NATS is down.

Only NATS connects lazily; the other coordinator backends still fail at
boot as before.

cluster::coordinator::coordinator_reconnect_tests starts a node against a
NATS port with nothing behind it, checks it boots with a coordinator and
reports it disconnected, subscribes, then starts NATS on that port: the
node connects on its own and the subscription receives a message from a
second client. Stopping and restarting NATS shows disconnected, then
connected, and the same subscription keeps working.

.gitea/workflows/publish.yml

@@ -14,11 +14,8 @@
 #     crates/types/src/branding.rs, not Cargo.toml, and the image is tagged
 #     with it, so a tag beside an unbumped macro would publish an image that
 #     reports a different version from its tag.
-#   * the tag must be on main or on a release/* branch, so an image never
+#   * the tag must be on main, so an image never describes code that was never
-#     describes code that was never reviewed onto one of them. A release/*
+#     reviewed onto the default branch.
-#     branch carries a hotfix: it starts at an earlier release tag, takes
-#     fixes through pull requests into it, and is tagged there, so production
-#     can get a fix without everything that has landed on main since.
 #
 # :latest moves with every published tag: tags are cut by the weekly release
 # (or by hand for a real release); there are no prerelease tags here.
@@ -60,13 +57,8 @@ jobs:
             echo "Refusing to publish an image that would report the wrong version." >&2
             exit 1
           fi
-          commit="$(git rev-parse "${TAG}^{commit}")"
+          git merge-base --is-ancestor "$(git rev-parse "${TAG}^{commit}")" origin/main \
-          on=""
+            || { echo "$TAG is not on main" >&2; exit 1; }
-          for ref in origin/main $(git for-each-ref --format='%(refname:short)' 'refs/remotes/origin/release/*'); do
-            if git merge-base --is-ancestor "$commit" "$ref"; then on="$ref"; break; fi
-          done
-          [ -n "$on" ] || { echo "$TAG is not on main or a release/* branch" >&2; exit 1; }
-          echo "$TAG is on $on"
           echo "version=$V" >> "$GITHUB_OUTPUT"
           echo "version $V"
 

crates/common/src/manager/backup.rs

@@ -23,6 +23,13 @@ use utils::{UnwrapFailure, codec::leb128::Leb128_};
 
 pub(super) const MAGIC_MARKER: u8 = 123;
 
+// inbuxa: blobs kept under a fixed name instead of a content hash. Nothing
+// links to them, so the export names them outright.
+const NAMED_BLOBS: &[&[u8]] = &[
+    crate::manager::SPAM_CLASSIFIER_KEY,
+    crate::manager::SPAM_TRAINER_KEY,
+];
+
 #[derive(Debug, Clone, Copy, Hash, PartialEq, Eq)]
 pub(super) enum Family {
     Data = 0,
@@ -143,15 +150,21 @@ impl Core {
                     .await
                     .failed("Failed to iterate over data store");
 
-                for hash in blobs {
+                // inbuxa: the trained spam classifier and its trainer state are
+                // blobs stored under fixed names with no blob link, so the walk
+                // over links above never reaches them.
+                let named = NAMED_BLOBS.iter().map(|key| key.to_vec());
+                for key in blobs
+                    .into_iter()
+                    .map(|hash| hash.as_slice().to_vec())
+                    .chain(named)
+                {
                     if let Some(blob) = blob_store
-                        .get_blob(hash.as_slice(), 0..usize::MAX)
+                        .get_blob(&key, 0..usize::MAX)
                         .await
                         .failed("Failed to get blob")
                     {
-                        writer
+                        writer.send((key, blob)).failed("Failed to send key");
-                            .send((hash.as_slice().to_vec(), blob))
-                            .failed("Failed to send key");
                     }
                 }
             }),
@@ -323,7 +336,13 @@ impl Family {
                 SUBSPACE_REGISTRY_IDX,
                 SUBSPACE_REGISTRY_PK,
                 SUBSPACE_DIRECTORY,
-                store::SUBSPACE_INBUXA, // inbuxa: masked email
+                // inbuxa: registry objects the upstream list left out, so an
+                // export dropped them: archived items (undelete) and spam
+                // training samples. Their indexes and id counters already
+                // travel in this family and in `data`, so they ride along.
+                SUBSPACE_DELETED_ITEMS,
+                SUBSPACE_SPAM_SAMPLES,
+                store::SUBSPACE_INBUXA, // inbuxa: the fork's own data (masked email, undelete, policies)
             ],
             Family::Changelog => &[SUBSPACE_LOGS],
             Family::Queue => &[SUBSPACE_QUEUE_MESSAGE, SUBSPACE_QUEUE_EVENT],

crates/common/src/manager/boot.rs

@@ -54,6 +54,13 @@ Options:
   -o, --console                    Open the store console
   -h, --help                       Print help
   -V, --version                    Print version
+
+An export holds everything in the data and blob stores except short-lived
+in-memory state (rate limits, locks, greylisting) and the full-text search
+index, which belongs to one search backend. An import into an empty store
+queues the index to be rebuilt when the server next starts. EXPORT_TYPES
+limits an export to some of: data, registry, blob, changelog, queue, report,
+telemetry, tasks.
 "#
 );
 
@@ -256,10 +263,10 @@ impl BootManager {
                 telemetry.enable();
 
                 // Parse settings and restore
-                Box::pin(Core::parse(&mut bootstrap, storage))
+                let core = Box::pin(Core::parse(&mut bootstrap, storage)).await;
-                    .await
+                let imported = core.restore(path).await;
-                    .restore(path)
+                // inbuxa: the search index isn't exported; rebuild it
-                    .await;
+                core.queue_reindex(&imported).await;
                 std::process::exit(0);
             }
             StoreOp::Console => {

crates/common/src/manager/restore.rs

@@ -9,15 +9,22 @@
 use super::backup::MAGIC_MARKER;
 use crate::{Core, DATABASE_SCHEMA_VERSION};
 use lz4_flex::frame::FrameDecoder;
-use registry::schema::enums::CompressionAlgo;
+use registry::{
+    schema::{
+        enums::{CompressionAlgo, TaskStoreMaintenanceType},
+        structs::{Task, TaskStatus, TaskStoreMaintenance},
+    },
+    types::EnumImpl,
+};
 use std::{
     fs::File,
     io::{BufReader, ErrorKind, Read},
     path::{Path, PathBuf},
 };
 use store::{
-    BlobStore, IterateParams, SUBSPACE_BLOBS, SUBSPACE_COUNTER, SUBSPACE_INDEXES, SUBSPACE_QUOTA,
+    BlobStore, IterateParams, SUBSPACE_BLOBS, SUBSPACE_COUNTER, SUBSPACE_INDEXES,
-    SUBSPACE_REGISTRY_PK, Store, U32_LEN,
+    SUBSPACE_PROPERTY, SUBSPACE_QUOTA, SUBSPACE_REGISTRY_PK, SUBSPACE_TELEMETRY_SPAN, Store,
+    U32_LEN,
     write::{
         AnyClass, AnyKey, BatchBuilder, ValueClass,
         key::{DeserializeBigEndian, is_node_id_key},
@@ -27,7 +34,9 @@ use types::{collection::Collection, field::Field};
 use utils::{UnwrapFailure, failed};
 
 impl Core {
-    pub async fn restore(&self, src: PathBuf) {
+    /// Imports an export into an empty store and returns the subspaces it
+    /// wrote. inbuxa: the caller hands them to [`Core::queue_reindex`].
+    pub async fn restore(&self, src: PathBuf) -> Vec<u8> {
         // Backup the core
         let paths = if src.is_dir() {
             let mut paths = Vec::new();
@@ -64,6 +73,13 @@ impl Core {
             std::process::exit(1);
         }
 
+        let mut imported = paths
+            .iter()
+            .map(|path| KeyValueReader::new(path).subspace)
+            .collect::<Vec<_>>();
+        imported.sort_unstable();
+        imported.dedup();
+
         let mut tasks = Vec::new();
         for path in paths {
             let storage = self.storage.clone();
@@ -76,6 +92,54 @@ impl Core {
         for task in tasks {
             task.await.failed("Failed to wait for task");
         }
+
+        imported
+    }
+
+    /// inbuxa: an export never carries the full-text index. It is built by
+    /// and for one search backend (the SQL stores index into their own
+    /// tables, the key-value stores into a subspace, external engines keep it
+    /// themselves), so it would be wrong or unreadable after a move to
+    /// another one. Instead, an import queues the same reindex tasks an
+    /// administrator can queue by hand (`reindexAccounts` and
+    /// `reindexTelemetry` store maintenance), and the server rebuilds the
+    /// index for whatever search store it is configured with once it starts.
+    pub async fn queue_reindex(&self, imported: &[u8]) -> Vec<TaskStoreMaintenanceType> {
+        let mut queued = Vec::new();
+        if imported.contains(&SUBSPACE_PROPERTY) {
+            queued.push(TaskStoreMaintenanceType::ReindexAccounts);
+        }
+        if imported.contains(&SUBSPACE_TELEMETRY_SPAN) {
+            queued.push(TaskStoreMaintenanceType::ReindexTelemetry);
+        }
+        if queued.is_empty() {
+            return queued;
+        }
+
+        let mut batch = BatchBuilder::new();
+        for maintenance_type in &queued {
+            batch.schedule_task(Task::StoreMaintenance(TaskStoreMaintenance {
+                maintenance_type: *maintenance_type,
+                status: TaskStatus::now(),
+                shard_index: None,
+            }));
+        }
+        self.storage
+            .data
+            .write(batch.build_all())
+            .await
+            .failed("Failed to queue the reindex tasks");
+
+        println!(
+            "Queued {} to rebuild the search index; it runs when the server starts.",
+            queued
+                .iter()
+                .map(|t| t.as_str())
+                .collect::<Vec<_>>()
+                .join(" and ")
+        );
+
+        queued
     }
 }
 
@@ -125,17 +189,22 @@ async fn restore_file(store: Store, blob_store: BlobStore, path: &Path) {
         }
         SUBSPACE_COUNTER | SUBSPACE_QUOTA => {
             while let Some((key, value)) = reader.next() {
-                batch.add(
+                let class = ValueClass::Any(AnyClass {
-                    ValueClass::Any(AnyClass {
                     subspace: reader.subspace,
                     key,
-                    }),
+                });
-                    u64::from_le_bytes(
+                let value = u64::from_le_bytes(
                     value
                         .try_into()
                         .expect("Failed to deserialize counter/quota"),
-                    ) as i64,
+                ) as i64;
-                );
+                // inbuxa: the SQL stores add a negative amount with an UPDATE,
+                // which does nothing to a row that isn't there yet, so a
+                // negative counter vanished on import. Create the row first.
+                if value < 0 {
+                    batch.add(class.clone(), 0);
+                }
+                batch.add(class, value);
                 if batch.is_large_batch() {
                     store
                         .write(batch.build_all())

crates/coordinator/Cargo.toml

@@ -8,7 +8,7 @@ store = { path =  "../store" }
 registry = { path =  "../registry" }
 trc = { path = "../trc" }
 futures = { version = "0.3", optional = true }
-tokio = { version = "1.53", features = ["sync", "fs", "io-util"] }
+tokio = { version = "1.53", features = ["sync", "fs", "io-util", "rt", "time"] }
 async-nats = { version = "0.50", default-features = false, features = ["server_2_10", "server_2_11", "aws-lc-rs"], optional = true }
 zenoh = { version = "1.10.0", default-features = false, features = ["auth_pubkey", "transport_multilink", "transport_compression", "transport_quic", "transport_tcp", "transport_tls", "transport_udp"], optional = true }
 rdkafka = { version = "0.39", features = ["cmake-build"], optional = true }

crates/coordinator/src/backend/nats/mod.rs

@@ -2,13 +2,22 @@
  * SPDX-FileCopyrightText: 2020 Stalwart Labs LLC <[email protected]>
  *
  * SPDX-License-Identifier: AGPL-3.0-only OR LicenseRef-SEL
+ *
+ * Modified by Coffey Labs in 2026 for INBUXA.
  */
 
-use std::sync::Arc;
+use std::{
+    sync::{
+        Arc,
+        atomic::{AtomicBool, Ordering},
+    },
+    time::Duration,
+};
 
 use crate::Coordinator;
 use async_nats::Client;
 use registry::schema::structs::NatsCoordinator;
+use trc::ClusterEvent;
 
 pub mod pubsub;
 
@@ -47,9 +56,116 @@ impl NatsPubSub {
             opts = opts.token(credentials);
         }
 
+        // inbuxa: connect in the background and keep trying, so a node that
+        // starts while NATS is down still joins the cluster once NATS is
+        // back, instead of running without a coordinator until restarted;
+        // and report the connection going and coming back
+        let reporter = Arc::new(Reporter::default());
+        opts = opts.retry_on_initial_connect().event_callback({
+            let reporter = reporter.clone();
+            move |event| {
+                let reporter = reporter.clone();
+                async move { reporter.report(event) }
+            }
+        });
+        let connection_timeout = config.timeout_connection.into_inner();
+
         async_nats::connect_with_options(config.addresses.into_inner(), opts)
             .await
-            .map(|client| Coordinator::Nats(Arc::new(NatsPubSub { client })))
+            .map(|client| {
+                reporter.watch_first_connection(client.clone(), connection_timeout);
+                Coordinator::Nats(Arc::new(NatsPubSub { client }))
+            })
             .map_err(|err| format!("Failed to connect to Nats: {}", err))
     }
+
+    /// inbuxa: whether the client is connected to a NATS server right now.
+    pub fn is_connected(&self) -> bool {
+        matches!(
+            self.client.connection_state(),
+            async_nats::connection::State::Connected
+        )
+    }
+}
+
+/// inbuxa: reports the client's connection events as the server's own.
+#[derive(Default)]
+struct Reporter {
+    connected_once: AtomicBool,
+    // A failed attempt raises an error each time the client retries, every
+    // few seconds while NATS is down: report the first after each change
+    error_reported: AtomicBool,
+}
+
+impl Reporter {
+    fn report(&self, event: async_nats::Event) {
+        match event {
+            async_nats::Event::Connected => {
+                self.connected_once.store(true, Ordering::Relaxed);
+                self.error_reported.store(false, Ordering::Relaxed);
+                trc::event!(Cluster(ClusterEvent::CoordinatorConnected), Type = "nats");
+            }
+            async_nats::Event::Disconnected => {
+                self.error_reported.store(false, Ordering::Relaxed);
+                trc::event!(
+                    Cluster(ClusterEvent::CoordinatorDisconnected),
+                    Type = "nats",
+                    Details = "Connection lost; reconnecting in the background",
+                );
+            }
+            async_nats::Event::Closed => {
+                trc::event!(
+                    Cluster(ClusterEvent::CoordinatorDisconnected),
+                    Type = "nats",
+                    Details = "Connection closed; no further attempts will be made",
+                );
+            }
+            async_nats::Event::ClientError(async_nats::ClientError::MaxReconnects) => {
+                trc::event!(
+                    Cluster(ClusterEvent::CoordinatorDisconnected),
+                    Type = "nats",
+                    Details = "Gave up reconnecting (maxReconnects reached)",
+                );
+            }
+            async_nats::Event::ClientError(err) => {
+                if !self.error_reported.swap(true, Ordering::Relaxed) {
+                    trc::event!(
+                        Cluster(ClusterEvent::CoordinatorError),
+                        Type = "nats",
+                        Details = "Connection attempt failed; retrying",
+                        Reason = err.to_string(),
+                    );
+                }
+            }
+            event => {
+                trc::event!(
+                    Cluster(ClusterEvent::CoordinatorError),
+                    Type = "nats",
+                    Details = event.to_string(),
+                );
+            }
+        }
+    }
+
+    /// The first connection is made in the background, so say so when it
+    /// hasn't been made within the connection timeout. The client keeps
+    /// trying, and reports the connection when it comes.
+    fn watch_first_connection(self: &Arc<Self>, client: Client, timeout: Duration) {
+        let reporter = self.clone();
+        tokio::spawn(async move {
+            tokio::time::sleep(timeout).await;
+            if !reporter.connected_once.load(Ordering::Relaxed)
+                && !matches!(
+                    client.connection_state(),
+                    async_nats::connection::State::Connected
+                )
+            {
+                trc::event!(
+                    Cluster(ClusterEvent::CoordinatorDisconnected),
+                    Type = "nats",
+                    Details = "Not connected at startup; retrying in the background",
+                );
+            }
+        });
+    }
 }

crates/coordinator/src/dispatch.rs

@@ -2,6 +2,8 @@
  * SPDX-FileCopyrightText: 2020 Stalwart Labs LLC <[email protected]>
  *
  * SPDX-License-Identifier: AGPL-3.0-only OR LicenseRef-SEL
+ *
+ * Modified by Coffey Labs in 2026 for INBUXA.
  */
 
 use crate::{Coordinator, Msg, PubSubStream};
@@ -43,6 +45,17 @@ impl Coordinator {
     pub fn is_none(&self) -> bool {
         matches!(self, Coordinator::None)
     }
+
+    /// inbuxa: whether the coordinator is connected right now, for the
+    /// backends that track it (NATS); `None` for the others and when no
+    /// coordinator is configured.
+    pub fn is_connected(&self) -> Option<bool> {
+        match self {
+            #[cfg(feature = "nats")]
+            Coordinator::Nats(store) => Some(store.is_connected()),
+            _ => None,
+        }
+    }
 }
 
 impl PubSubStream {

crates/http/src/request.rs

@@ -562,6 +562,27 @@ impl ParseHttp for Server {
                         })
                         .into_http_response());
                     }
+                    // inbuxa: the cluster coordinator's connection, for
+                    // monitoring. It stays out of live and ready on purpose:
+                    // a node without its coordinator still serves mail, and
+                    // failing those would have an orchestrator restart, or
+                    // take out of service, every node at once when the
+                    // coordinator goes down
+                    "cluster" => {
+                        let coordinator = &self.core.storage.coordinator;
+                        let (status, state) = match coordinator.is_connected() {
+                            Some(true) => (StatusCode::OK, "connected"),
+                            Some(false) => (StatusCode::SERVICE_UNAVAILABLE, "disconnected"),
+                            None if coordinator.is_none() => (StatusCode::OK, "none"),
+                            None => (StatusCode::OK, "unknown"),
+                        };
+                        return Ok(http_proto::JsonResponse::with_status(
+                            status,
+                            serde_json::json!({ "coordinator": state }),
+                        )
+                        .no_cache()
+                        .into_http_response());
+                    }
                     _ => (),
                 }
             }

crates/jmap/src/inbuxa/telemetry.rs

@@ -427,9 +427,24 @@ pub(crate) async fn trace_query(
             }
             None => false,
         },
-        Property::QueueId => match value.as_str() {
+        // The queue id column is an integer on every search backend, and
+        // holds a trace's first queue id; the keywords carry all of them
+        Property::QueueId => match value
+            .as_str()
+            .and_then(|v| v.trim().parse::<u64>().ok())
+            .or_else(|| value.as_u64())
+        {
             Some(queue_id) => {
-                search.push(SearchFilter::eq(TracingSearchField::QueueId, queue_id.to_string()));
+                search.extend([
+                    SearchFilter::Or,
+                    SearchFilter::eq(TracingSearchField::QueueId, queue_id),
+                    SearchFilter::has_text(
+                        TracingSearchField::Keywords,
+                        queue_id.to_string(),
+                        nlp::language::Language::None,
+                    ),
+                    SearchFilter::End,
+                ]);
                 true
             }
             None => false,

crates/jmap/src/registry/mapping/report.rs

@@ -2,8 +2,6 @@
  * SPDX-FileCopyrightText: 2020 Stalwart Labs LLC <[email protected]>
  *
  * SPDX-License-Identifier: AGPL-3.0-only OR LicenseRef-SEL
- *
- * Modified by Coffey Labs in 2026 for INBUXA.
  */
 
 use crate::{
@@ -17,42 +15,22 @@ use jmap_proto::{error::set::SetError, types::state::State};
 use jmap_tools::{Key, Value};
 use registry::{
     jmap::IntoValue,
-    schema::{
+    schema::prelude::{Object, ObjectInner, ObjectType, Property},
-        prelude::{Object, ObjectInner, ObjectType, Property},
-        structs::Task,
-    },
     types::{EnumImpl, datetime::UTCDateTime},
 };
-use services::task_manager::lock::TaskLockManager;
 use smtp::reporting::index::{ExternalReportIndex, InternalReportIndex};
 use std::str::FromStr;
 use store::{
     U64_LEN, ValueKey,
     registry::{RegistryFilter, RegistryFilterValue, RegistryQuery},
-    write::{BatchBuilder, RegistryClass, TaskQueueClass, ValueClass, key::KeySerializer},
+    write::{BatchBuilder, RegistryClass, ValueClass, key::KeySerializer},
 };
 use trc::AddContext;
 use types::id::Id;
 
 pub(crate) async fn report_set(
-    set: RegistrySetResponse<'_>,
+    mut set: RegistrySetResponse<'_>,
 ) -> trc::Result<RegistrySetResponse<'_>> {
-    // inbuxa: task locks taken to reschedule reports are released however
-    // the request ends; a held lock is renewed, so a leaked one would keep
-    // the report's task from ever running
-    let server = set.server;
-    let mut locked_tasks = Vec::new();
-    let result = report_set_locked(set, &mut locked_tasks).await;
-    for task_id in locked_tasks {
-        server.remove_index_lock(task_id).await;
-    }
-    result
-}
-
-async fn report_set_locked<'x>(
-    mut set: RegistrySetResponse<'x>,
-    locked_tasks: &mut Vec<u64>,
-) -> trc::Result<RegistrySetResponse<'x>> {
     let object_id = set.object_type.to_id();
 
     // Reports cannot be created
@@ -111,45 +89,12 @@ async fn report_set_locked<'x>(
                 .get_value::<Object>(ValueKey::from(key.clone()))
                 .await?
             {
-                // inbuxa: the report's task shares its id. Hold the task
-                // while its queue rows move, as x:Task/set does, and move the
-                // row the task is actually queued under
-                if !set.server.try_lock_task(item_id).await {
-                    set.response.not_updated.append(
-                        id,
-                        SetError::forbidden().with_description(
-                            "The report is being sent and cannot be rescheduled".to_string(),
-                        ),
-                    );
-                    continue;
-                }
-                locked_tasks.push(item_id);
-                let queued = set
-                    .server
-                    .store()
-                    .get_value::<Task>(ValueKey::from(ValueClass::TaskQueue(
-                        TaskQueueClass::Task { id: item_id },
-                    )))
-                    .await?;
-
                 match &mut report_obj.inner {
                     ObjectInner::DmarcInternalReport(report) => {
-                        report.reschedule_ops(
+                        report.reschedule_ops(&mut batch, item_id, report_obj.revision, deliver_at);
-                            &mut batch,
-                            item_id,
-                            report_obj.revision,
-                            deliver_at,
-                            queued.as_ref(),
-                        );
                     }
                     ObjectInner::TlsInternalReport(report) => {
-                        report.reschedule_ops(
+                        report.reschedule_ops(&mut batch, item_id, report_obj.revision, deliver_at);
-                            &mut batch,
-                            item_id,
-                            report_obj.revision,
-                            deliver_at,
-                            queued.as_ref(),
-                        );
                     }
                     _ => {}
                 }
@@ -211,9 +156,6 @@ async fn report_set_locked<'x>(
             .write(batch.build_all())
             .await
             .caused_by(trc::location!())?;
-        // inbuxa: a rescheduled report may now be due sooner than the task
-        // manager's next scan
-        set.server.notify_task_queue();
     }
 
     Ok(set)

crates/jmap/src/registry/mapping/task.rs

@@ -463,10 +463,15 @@ pub(crate) async fn task_query(
                 .set_values(typ.is_some()),
             |key, value| {
                 if let Some(typ) = typ {
-                    // inbuxa: a row whose type can't be read matches no type
+                    let task_type =
-                    // filter; the task manager logs and repairs it
+                        TaskType::from_id(value.deserialize_be_u16(0)?).ok_or_else(|| {
-                    let task_type = value.deserialize_be_u16(0).ok().and_then(TaskType::from_id);
+                            trc::StoreEvent::DataCorruption
-                    if task_type != Some(typ) {
+                                .into_err()
+                                .ctx(trc::Key::Key, key.to_vec())
+                                .ctx(trc::Key::Value, value.to_vec())
+                                .caused_by(trc::location!())
+                        })?;
+                    if task_type != typ {
                         return Ok(true);
                     }
                 }

crates/services/src/broadcast/subscriber.rs

@@ -26,7 +26,7 @@ pub fn spawn_broadcast_subscriber(inner: Arc<Inner>, mut shutdown_rx: watch::Rec
     };
 
     tokio::spawn(async move {
-        let mut retry_count = 0;
+        let mut retry_count: u32 = 0;
 
         trc::event!(Cluster(ClusterEvent::SubscriberStart));
 
@@ -53,7 +53,7 @@ pub fn spawn_broadcast_subscriber(inner: Arc<Inner>, mut shutdown_rx: watch::Rec
                     );
 
                     match tokio::time::timeout(
-                        Duration::from_secs(1 << retry_count.max(6)),
+                        subscribe_retry_delay(retry_count),
                         shutdown_rx.changed(),
                     )
                     .await
@@ -62,7 +62,7 @@ pub fn spawn_broadcast_subscriber(inner: Arc<Inner>, mut shutdown_rx: watch::Rec
                             break;
                         }
                         Err(_) => {
-                            retry_count += 1;
+                            retry_count = retry_count.saturating_add(1);
                             continue;
                         }
                     }
@@ -234,6 +234,11 @@ pub fn spawn_broadcast_subscriber(inner: Arc<Inner>, mut shutdown_rx: watch::Rec
     });
 }
 
+/// Delay before the next subscribe attempt: 1 s, 2 s, 4 s ... capped at 64 s.
+fn subscribe_retry_delay(retry_count: u32) -> Duration {
+    Duration::from_secs(1u64 << retry_count.min(6))
+}
+
 fn log_event(event: &BroadcastEvent) -> trc::Value {
     match event {
         BroadcastEvent::PushNotification(notification) => match notification {
@@ -296,3 +301,19 @@ fn log_event(event: &BroadcastEvent) -> trc::Value {
         BroadcastEvent::QueueRefresh => "QueueRefresh".into(),
     }
 }
+
+#[cfg(test)]
+mod tests {
+    use super::subscribe_retry_delay;
+    use std::time::Duration;
+
+    #[test]
+    fn subscribe_retry_backoff_grows_then_caps() {
+        let schedule: Vec<u64> = (0..10)
+            .map(|n| subscribe_retry_delay(n).as_secs())
+            .collect();
+        assert_eq!(schedule, vec![1, 2, 4, 8, 16, 32, 64, 64, 64, 64]);
+        // No shift overflow at the top of the range.
+        assert_eq!(subscribe_retry_delay(u32::MAX), Duration::from_secs(64));
+    }
+}

crates/services/src/task_manager/index.rs

@@ -567,20 +567,14 @@ async fn build_contact_document(
 }
 
 
-// inbuxa: MON-16: a trace's search document, when trace search is on:
+// inbuxa: MON-16: a trace's search document, when trace search is on
-// its event types, queue ids, and addresses, their domains, hosts, IPs,
-// message ids and account names as keywords
 async fn build_tracing_span_document(
     server: &Server,
     span_id: u64,
 ) -> trc::Result<Option<IndexDocument>> {
     use common::telemetry::tracers::store::MaybeTrace;
-    use registry::schema::{enums::SearchTracingField, structs::Search};
+    use registry::schema::structs::Search;
-    use store::{
+    use store::write::{TelemetryClass, ValueClass};
-        search::TracingSearchField,
-        write::{TelemetryClass, ValueClass},
-    };
-    use trc::Key;
 
     let settings = server
         .registry()
@@ -590,7 +584,6 @@ async fn build_tracing_span_document(
     if !settings.index_telemetry {
         return Ok(None);
     }
-    let wants = |field: SearchTracingField| settings.index_tracing_fields.iter().any(|f| *f == field);
     let Some(MaybeTrace(Some(trace))) = server
         .tracing_store()
         .get_value::<MaybeTrace>(ValueKey::from(ValueClass::Telemetry(TelemetryClass::Span(
@@ -601,23 +594,67 @@ async fn build_tracing_span_document(
         return Ok(None);
     };
 
-    let mut document = IndexDocument::new(SearchIndex::Tracing).with_id(span_id);
+    Ok(Some(trace_search_document(
-    let mut seen = store::ahash::AHashSet::new();
+        span_id,
-    for event in trace.events.iter() {
+        &trace,
-        if wants(SearchTracingField::EventType) && seen.insert(event.event.as_str().to_string()) {
+        &settings
-            document.index_keyword(TracingSearchField::EventType, event.event.as_str());
+            .index_tracing_fields
+            .iter()
+            .copied()
+            .collect::<Vec<_>>(),
+    )))
 }
+
+/// inbuxa: MON-16: the search document for a stored trace.
+///
+/// The event type and queue id columns are integers on every search backend
+/// (BIGINT on PostgreSQL and MySQL, long on Elasticsearch), and each holds a
+/// single value per trace: the event type is the trace's opening event, the
+/// one `x:Trace/query` filters on, and the queue id is the first queue id the
+/// trace mentions. Every queue id also goes into the keywords, so a session
+/// that queued several messages is found by any of them.
+pub fn trace_search_document(
+    span_id: u64,
+    trace: &registry::schema::structs::Trace,
+    fields: &[registry::schema::enums::SearchTracingField],
+) -> IndexDocument {
+    use registry::schema::{enums::SearchTracingField, structs::TraceValue};
+    use store::search::TracingSearchField;
+    use trc::Key;
+
+    let wants = |field: SearchTracingField| fields.contains(&field);
+    let mut document = IndexDocument::new(SearchIndex::Tracing).with_id(span_id);
+    if wants(SearchTracingField::EventType)
+        && let Some(first) = trace.events.iter().next()
+    {
+        document.index_unsigned(TracingSearchField::EventType, first.event.to_id() as u64);
+    }
+
+    let mut seen = store::ahash::AHashSet::new();
+    let mut queue_id_indexed = false;
+    for event in trace.events.iter() {
         for kv in event.key_values.iter() {
             let text = match &kv.value {
-                registry::schema::structs::TraceValue::String(v) => v.value.clone(),
+                TraceValue::String(v) => v.value.clone(),
-                registry::schema::structs::TraceValue::UnsignedInt(v) => v.value.to_string(),
+                TraceValue::UnsignedInt(v) => v.value.to_string(),
-                registry::schema::structs::TraceValue::IpAddr(v) => v.value.to_string(),
+                TraceValue::IpAddr(v) => v.value.to_string(),
                 _ => continue,
             };
             match kv.key {
-                Key::QueueId if wants(SearchTracingField::QueueId) => {
+                Key::QueueId => {
-                    if seen.insert(format!("q:{text}")) {
+                    let Ok(queue_id) = text.parse::<u64>() else {
-                        document.index_keyword(TracingSearchField::QueueId, &text);
+                        continue;
+                    };
+                    if wants(SearchTracingField::QueueId) && !queue_id_indexed {
+                        document.index_unsigned(TracingSearchField::QueueId, queue_id);
+                        queue_id_indexed = true;
+                    }
+                    if wants(SearchTracingField::Keywords) && seen.insert(format!("k:{text}")) {
+                        document.index_text(
+                            TracingSearchField::Keywords,
+                            &text,
+                            nlp::language::Language::None,
+                        );
                     }
                 }
                 Key::From
@@ -648,7 +685,7 @@ async fn build_tracing_span_document(
             }
         }
     }
-    Ok(Some(document))
+    document
 }
 
 // inbuxa: UD-1, UD-4: archives a deleted file, event or contact noted at

crates/services/src/task_manager/manager.rs

@@ -2,8 +2,6 @@
  * SPDX-FileCopyrightText: 2020 Stalwart Labs LLC <[email protected]>
  *
  * SPDX-License-Identifier: AGPL-3.0-only OR LicenseRef-SEL
- *
- * Modified by Coffey Labs in 2026 for INBUXA.
  */
 
 use crate::task_manager::acme::AcmeTask;
@@ -29,7 +27,6 @@ use common::network::limiter::ConcurrencyLimiter;
 use common::network::{ServerInstance, TcpAcceptor};
 use common::{Inner, Server};
 use registry::schema::enums::TaskType;
-use registry::schema::prelude::ObjectType;
 use registry::schema::structs::{
     Task, TaskManager, TaskRetryStrategy, TaskStatus, TaskStatusFailed, TaskStatusRetry,
 };
@@ -340,7 +337,6 @@ impl TaskQueueManager for Server {
 
         // Retrieve tasks pending to be processed
         let mut tasks = Vec::new();
-        let mut unreadable = Vec::new();
         let now = Instant::now();
         let mut next_event = None;
         let roles = &self.core.network.roles;
@@ -355,21 +351,12 @@ impl TaskQueueManager for Server {
                         let task_id = key.deserialize_be_u64(U64_LEN)?;
 
                         if task_due <= now_timestamp {
-                            // inbuxa: a row whose task type can't be read is
+                            let task_type_idx = value.deserialize_be_u16(0)?;
-                            // set aside, not allowed to end the scan: every
+                            let task_type = TaskType::from_id(task_type_idx).ok_or_else(|| {
-                            // task due after it would wait behind it
+                                trc::StoreEvent::DataCorruption
-                            let Some((task_type_idx, task_type)) = value
+                                    .caused_by(trc::location!())
-                                .deserialize_be_u16(0)
+                                    .ctx(trc::Key::Value, value)
-                                .ok()
+                            })?;
-                                .and_then(|idx| TaskType::from_id(idx).map(|typ| (idx, typ)))
-                            else {
-                                unreadable.push(UnreadableDueRow {
-                                    due: task_due,
-                                    id: task_id,
-                                    value: value.to_vec(),
-                                });
-                                return Ok(true);
-                            };
                             let enabled = match task_type {
                                 TaskType::IndexDocument
                                 | TaskType::UnindexDocument
@@ -459,11 +446,6 @@ impl TaskQueueManager for Server {
                 );
             });
 
-        if !unreadable.is_empty() && repair_due_rows(self, unreadable).await {
-            // Look again at once for the rows that were rewritten
-            self.notify_task_queue();
-        }
-
         if !tasks.is_empty() {
             trc::event!(
                 TaskManager(TaskManagerEvent::TaskAcquired),
@@ -704,114 +686,3 @@ impl TaskResult {
         )
     }
 }
-
-/// inbuxa: a task queue row whose task type could not be read.
-struct UnreadableDueRow {
-    due: u64,
-    id: u64,
-    value: Vec<u8>,
-}
-
-/// inbuxa: logs each unreadable queue row and repairs it from the task it
-/// schedules. The task row says what the task is, so the queue row is
-/// rewritten with that task's type; a row with no task behind it is removed.
-///
-/// Rescheduling an internal DMARC or TLS report wrote the report's object
-/// type into the queue row instead of the task type. Such a row is the time
-/// an administrator chose, so the task is moved to it as the reschedule
-/// meant to do: the task row takes that due, and a queue row left at the
-/// task's previous due is removed. Returns whether any row was repaired.
-async fn repair_due_rows(server: &Server, rows: Vec<UnreadableDueRow>) -> bool {
-    let mut repaired = false;
-    for row in rows {
-        let UnreadableDueRow { due, id, value } = row;
-        trc::error!(
-            trc::StoreEvent::DataCorruption
-                .into_err()
-                .id(id)
-                .ctx(trc::Key::Due, trc::Value::Timestamp(due))
-                .ctx(
-                    trc::Key::Key,
-                    [due.to_be_bytes(), id.to_be_bytes()].concat()
-                )
-                .ctx(trc::Key::Value, value.clone())
-                .details("Unreadable task queue row skipped")
-                .caused_by(trc::location!())
-        );
-
-        let task_key = ValueClass::TaskQueue(TaskQueueClass::Task { id });
-        let due_key = ValueClass::TaskQueue(TaskQueueClass::Due { id, due });
-        let task = match server
-            .store()
-            .get_value::<Task>(ValueKey::from(task_key.clone()))
-            .await
-        {
-            Ok(task) => task,
-            Err(err) => {
-                trc::error!(
-                    err.id(id)
-                        .details("Failed to read the task of an unreadable queue row.")
-                        .caused_by(trc::location!())
-                );
-                continue;
-            }
-        };
-
-        let mut batch = BatchBuilder::new();
-        let action = if let Some(mut task) = task {
-            let task_type = task.object_type();
-            batch.assert_value(task_key.clone(), AssertValue::Some);
-            if rescheduled_report_type(&value) == Some(task_type) {
-                let old_due = task.due_timestamp();
-                if old_due != due {
-                    batch.clear(ValueClass::TaskQueue(TaskQueueClass::Due {
-                        id,
-                        due: old_due,
-                    }));
-                }
-                task.set_status(TaskStatus::at(due as i64));
-            }
-            batch
-                .set(due_key, task_type.to_id().serialize())
-                .set(task_key, task.to_pickled_vec());
-            "Rewrote the queue row from its task."
-        } else {
-            batch.clear(due_key);
-            "Removed a queue row with no task."
-        };
-
-        match server.store().write(batch.build_all()).await {
-            Ok(_) => {
-                repaired = true;
-                trc::event!(
-                    TaskManager(TaskManagerEvent::TaskIgnored),
-                    Id = id,
-                    Due = trc::Value::Timestamp(due),
-                    Reason = action,
-                );
-            }
-            Err(err) if err.matches(trc::EventType::Store(trc::StoreEvent::AssertValueFailed)) => {
-                // The task went away meanwhile; the next scan looks again
-            }
-            Err(err) => {
-                trc::error!(
-                    err.id(id)
-                        .details("Failed to repair an unreadable queue row.")
-                        .caused_by(trc::location!())
-                );
-            }
-        }
-    }
-    repaired
-}
-
-/// inbuxa: the task type a report reschedule meant, when a queue row holds
-/// an internal report's object type (the value that reschedule wrote).
-fn rescheduled_report_type(value: &[u8]) -> Option<TaskType> {
-    let id = u16::from_be_bytes(value.get(..2)?.try_into().ok()?);
-    match ObjectType::from_id(id)? {
-        ObjectType::DmarcInternalReport => Some(TaskType::DmarcReport),
-        ObjectType::TlsInternalReport => Some(TaskType::TlsReport),
-        _ => None,
-    }
-}

crates/smtp/src/reporting/index.rs

@@ -2,8 +2,6 @@
  * SPDX-FileCopyrightText: 2020 Stalwart Labs LLC <[email protected]>
  *
  * SPDX-License-Identifier: AGPL-3.0-only OR LicenseRef-SEL
- *
- * Modified by Coffey Labs in 2026 for INBUXA.
  */
 
 use registry::{
@@ -42,49 +40,35 @@ pub trait InternalReportIndex: ObjectImpl {
 
     fn primary_key(&self) -> ValueClass;
 
-    /// Moves the report's delivery, and its queued task, to `at`.
-    ///
-    /// inbuxa: the new queue row carries the task's type, as
-    /// `schedule_task_with_id` writes it, and the task row gets the new due
-    /// too. `queued` is the task as stored: its due, not the report's
-    /// `deliverAt`, is the queue row that exists (they differ once the task
-    /// has been retried).
     fn reschedule_ops(
         &mut self,
         batch: &mut BatchBuilder,
         item_id: u64,
         revision: u64,
         at: UTCDateTime,
-        queued: Option<&Task>,
     ) {
         let current_deliver_at = self.deliver_at();
-        let current_due = current_deliver_at.timestamp() as u64;
-        let queued_due = queued.map_or(current_due, |task| task.due_timestamp());
-        let new_due = at.timestamp() as u64;
 
-        if current_deliver_at != at || queued_due != new_due {
+        if current_deliver_at != at {
             let object = Self::OBJECT;
             let object_id = object.to_id();
             let key = ValueClass::Registry(RegistryClass::Item { object_id, item_id });
 
             self.set_deliver_at(at);
 
-            batch.assert_value(key.clone(), AssertValue::Hash(revision));
-            if queued_due != new_due {
-                batch.clear(ValueClass::TaskQueue(TaskQueueClass::Due {
-                    id: item_id,
-                    due: queued_due,
-                }));
-            }
-            // A row an earlier reschedule left at the report's deliverAt
-            if current_due != new_due && current_due != queued_due {
-                batch.clear(ValueClass::TaskQueue(TaskQueueClass::Due {
-                    id: item_id,
-                    due: current_due,
-                }));
-            }
             batch
-                .schedule_task_with_id(item_id, self.task(item_id))
+                .assert_value(key.clone(), AssertValue::Hash(revision))
+                .clear(ValueClass::TaskQueue(TaskQueueClass::Due {
+                    id: item_id,
+                    due: current_deliver_at.timestamp() as u64,
+                }))
+                .set(
+                    ValueClass::TaskQueue(TaskQueueClass::Due {
+                        id: item_id,
+                        due: at.timestamp() as u64,
+                    }),
+                    object_id.serialize(),
+                )
                 .set(key, self.to_pickled_vec());
         }
     }

crates/trc/src/event/enums.rs

@@ -10,8 +10,9 @@
 
 // inbuxa: 637 to 641 are the fork's SCIM events (SCIM-54); 642 is
 // auth.legacy-protocol-refused (legacy-protocols LP-6); 643 is
-// security.legacy-protocols-changed (LP-8)
+// security.legacy-protocols-changed (LP-8); 644 to 646 are the cluster
-pub const TOTAL_EVENT_COUNT: usize = 644;
+// coordinator's connection events
+pub const TOTAL_EVENT_COUNT: usize = 647;
 pub const TOTAL_METRIC_COUNT: usize = 369;
 
 #[derive(Debug, Clone, Copy, PartialEq, Eq, Hash)]
@@ -150,6 +151,10 @@ pub enum ClusterEvent {
     MessageSkipped = 47,
     MessageInvalid = 49,
     NodeIdRenewed = 275,
+    // inbuxa: the coordinator's connection
+    CoordinatorConnected = 644,
+    CoordinatorDisconnected = 645,
+    CoordinatorError = 646,
 }
 
 #[derive(Debug, Clone, Copy, PartialEq, Eq, Hash)]

crates/trc/src/event/enums_impl.rs

@@ -81,6 +81,10 @@ impl EventType {
             b"cluster.message-skipped" => EventType::Cluster(ClusterEvent::MessageSkipped),
             b"cluster.message-invalid" => EventType::Cluster(ClusterEvent::MessageInvalid),
             b"cluster.node-id-renewed" => EventType::Cluster(ClusterEvent::NodeIdRenewed),
+            // inbuxa: coordinator connection
+            b"cluster.coordinator-connected" => EventType::Cluster(ClusterEvent::CoordinatorConnected),
+            b"cluster.coordinator-disconnected" => EventType::Cluster(ClusterEvent::CoordinatorDisconnected),
+            b"cluster.coordinator-error" => EventType::Cluster(ClusterEvent::CoordinatorError),
             b"dane.authentication-success" => EventType::Dane(DaneEvent::AuthenticationSuccess),
             b"dane.authentication-failure" => EventType::Dane(DaneEvent::AuthenticationFailure),
             b"dane.no-certificates-found" => EventType::Dane(DaneEvent::NoCertificatesFound),
@@ -742,6 +746,14 @@ impl EventType {
             EventType::Cluster(ClusterEvent::MessageSkipped) => "cluster.message-skipped",
             EventType::Cluster(ClusterEvent::MessageInvalid) => "cluster.message-invalid",
             EventType::Cluster(ClusterEvent::NodeIdRenewed) => "cluster.node-id-renewed",
+            // inbuxa: coordinator connection
+            EventType::Cluster(ClusterEvent::CoordinatorConnected) => {
+                "cluster.coordinator-connected"
+            }
+            EventType::Cluster(ClusterEvent::CoordinatorDisconnected) => {
+                "cluster.coordinator-disconnected"
+            }
+            EventType::Cluster(ClusterEvent::CoordinatorError) => "cluster.coordinator-error",
             EventType::Dane(DaneEvent::AuthenticationSuccess) => "dane.authentication-success",
             EventType::Dane(DaneEvent::AuthenticationFailure) => "dane.authentication-failure",
             EventType::Dane(DaneEvent::NoCertificatesFound) => "dane.no-certificates-found",
@@ -1524,6 +1536,10 @@ impl EventType {
             EventType::Cluster(ClusterEvent::MessageSkipped) => 47,
             EventType::Cluster(ClusterEvent::MessageInvalid) => 49,
             EventType::Cluster(ClusterEvent::NodeIdRenewed) => 275,
+            // inbuxa: coordinator connection
+            EventType::Cluster(ClusterEvent::CoordinatorConnected) => 644,
+            EventType::Cluster(ClusterEvent::CoordinatorDisconnected) => 645,
+            EventType::Cluster(ClusterEvent::CoordinatorError) => 646,
             EventType::Dane(DaneEvent::AuthenticationSuccess) => 67,
             EventType::Dane(DaneEvent::AuthenticationFailure) => 66,
             EventType::Dane(DaneEvent::NoCertificatesFound) => 69,
@@ -2176,6 +2192,10 @@ impl EventType {
             47 => Some(EventType::Cluster(ClusterEvent::MessageSkipped)),
             49 => Some(EventType::Cluster(ClusterEvent::MessageInvalid)),
             275 => Some(EventType::Cluster(ClusterEvent::NodeIdRenewed)),
+            // inbuxa: coordinator connection
+            644 => Some(EventType::Cluster(ClusterEvent::CoordinatorConnected)),
+            645 => Some(EventType::Cluster(ClusterEvent::CoordinatorDisconnected)),
+            646 => Some(EventType::Cluster(ClusterEvent::CoordinatorError)),
             67 => Some(EventType::Dane(DaneEvent::AuthenticationSuccess)),
             66 => Some(EventType::Dane(DaneEvent::AuthenticationFailure)),
             69 => Some(EventType::Dane(DaneEvent::NoCertificatesFound)),
@@ -3114,6 +3134,10 @@ impl EventType {
             EventType::Auth(AuthEvent::TooManyAttempts) => Level::Warn,
             EventType::Calendar(CalendarEvent::AlarmFailed) => Level::Warn,
             EventType::Cluster(ClusterEvent::SubscriberDisconnected) => Level::Warn,
+            // inbuxa: coordinator connection
+            EventType::Cluster(ClusterEvent::CoordinatorConnected) => Level::Info,
+            EventType::Cluster(ClusterEvent::CoordinatorDisconnected) => Level::Warn,
+            EventType::Cluster(ClusterEvent::CoordinatorError) => Level::Warn,
             EventType::Delivery(DeliveryEvent::MissingOutboundHostname) => Level::Warn,
             EventType::Delivery(DeliveryEvent::ConcurrencyLimitExceeded) => Level::Warn,
             EventType::Delivery(DeliveryEvent::RateLimitExceeded) => Level::Warn,
@@ -3244,6 +3268,10 @@ impl EventType {
             EventType::Cluster(ClusterEvent::MessageSkipped) => "PubSub message skipped",
             EventType::Cluster(ClusterEvent::MessageInvalid) => "Invalid PubSub message",
             EventType::Cluster(ClusterEvent::NodeIdRenewed) => "Node ID renewed",
+            // inbuxa: coordinator connection
+            EventType::Cluster(ClusterEvent::CoordinatorConnected) => "Coordinator connected",
+            EventType::Cluster(ClusterEvent::CoordinatorDisconnected) => "Coordinator unavailable",
+            EventType::Cluster(ClusterEvent::CoordinatorError) => "Coordinator error",
             EventType::Dane(DaneEvent::AuthenticationSuccess) => "DANE authentication successful",
             EventType::Dane(DaneEvent::AuthenticationFailure) => "DANE authentication failed",
             EventType::Dane(DaneEvent::NoCertificatesFound) => "No certificates found for DANE",
@@ -4322,6 +4350,10 @@ impl EventType {
             EventType::Cluster(ClusterEvent::MessageSkipped),
             EventType::Cluster(ClusterEvent::MessageInvalid),
             EventType::Cluster(ClusterEvent::NodeIdRenewed),
+            // inbuxa: coordinator connection
+            EventType::Cluster(ClusterEvent::CoordinatorConnected),
+            EventType::Cluster(ClusterEvent::CoordinatorDisconnected),
+            EventType::Cluster(ClusterEvent::CoordinatorError),
             EventType::Dane(DaneEvent::AuthenticationSuccess),
             EventType::Dane(DaneEvent::AuthenticationFailure),
             EventType::Dane(DaneEvent::NoCertificatesFound),

crates/types/src/branding.rs

@@ -81,7 +81,7 @@ fn legacy_setting(name: &str, is_set: impl Fn(&str) -> bool) -> Option<String> {
 #[macro_export]
 macro_rules! brand_version {
     () => {
-        "2026.9.24.4"
+        "2026.9.24.3"
     };
 }
 

docs/spec/features/monitoring.md

@@ -212,10 +212,15 @@ unchanged.
 - **MON-16.** With `indexTelemetry` on, storing a trace schedules an
   `IndexTrace` task. The task builds one document for `SearchIndex::Tracing`
   with the fields named in `indexTracingFields`:
-  - `eventType`: every event type in the trace;
+  - `eventType`: the trace's opening event, as its numeric id;
-  - `queueId`: every `queueId` value;
+  - `queueId`: the first `queueId` value, as an integer;
   - `keywords`: every address in `from` and `to`, each address's domain, every
-    `domain`, `hostname`, `remoteIp`, `messageId` and `accountName` value.
+    `domain`, `hostname`, `remoteIp`, `messageId` and `accountName` value,
+    and every `queueId` value.
+  The event type and queue id are single integer columns on every search
+  backend (BIGINT on PostgreSQL and MySQL), so the `queueId` filter matches
+  the column or any queue id in the keywords, and a session that queued
+  several messages is found by each of them.
   So searching `example.org` finds every trace to or from that domain, as the
   upstream suite expects. With `indexTelemetry` off nothing is indexed, and
   the `text` and `queueId` filters are refused (see "Interfaces").

resources/schema/schema.json.sha256

@@ -1 +1 @@
-VbnFuwCOTBh0s2T-NuRhb2JaJr8Jl5s3LgXv4Pv2sTg
+XFI3xuKC_rH1KZyaVBF0uTIiRDXRqyYboijquiGz2eg

tests/src/cluster/coordinator.rs

@@ -0,0 +1,145 @@
+/*
+ * SPDX-FileCopyrightText: 2026 Coffey Labs
+ *
+ * SPDX-License-Identifier: AGPL-3.0-only
+ */
+
+//! A node that starts while its NATS coordinator is down joins the cluster
+//! once NATS comes up, without a restart, and reports the coordinator's
+//! connection on `/healthz/cluster` as it goes and comes back.
+
+use crate::utils::server::TestServerBuilder;
+use coordinator::Coordinator;
+use registry::{
+    schema::{
+        enums::NetworkListenerProtocol,
+        structs::{Coordinator as CoordinatorSetting, NatsCoordinator},
+    },
+    types::map::Map,
+};
+use serde_json::{Value, json};
+use std::time::{Duration, Instant};
+use testcontainers::{
+    GenericImage, ImageExt, core::IntoContainerPort, core::WaitFor, runners::AsyncRunner,
+};
+
+const HTTP_PORT: u16 = 11_310;
+const TOPIC: &str = "inbuxa-coordinator-test";
+
+#[tokio::test(flavor = "multi_thread")]
+pub async fn coordinator_reconnect_tests() {
+    println!("Running coordinator reconnect tests...");
+
+    // A port with no NATS server behind it, yet
+    let nats_port = std::net::TcpListener::bind("127.0.0.1:0")
+        .unwrap()
+        .local_addr()
+        .unwrap()
+        .port();
+    let config = NatsCoordinator {
+        addresses: Map::new(vec![format!("127.0.0.1:{nats_port}")]),
+        use_tls: false,
+        timeout_connection: 1_000u64.into(),
+        ..Default::default()
+    };
+
+    // 1. The node starts, without a build error, while NATS is down, and
+    // says so
+    let test = TestServerBuilder::new("coordinator_reconnect_tests")
+        .await
+        .with_object(CoordinatorSetting::Nats(config.clone()))
+        .await
+        .with_listener(NetworkListenerProtocol::Http, "http", HTTP_PORT, true)
+        .await
+        .build()
+        .await;
+    let coordinator = test.server.core.storage.coordinator.clone();
+    assert!(
+        coordinator.is_enabled(),
+        "a coordinator, though not connected"
+    );
+    assert_eq!(coordinator.is_connected(), Some(false));
+    assert_eq!(
+        cluster_health().await,
+        (503, json!({"coordinator": "disconnected"}))
+    );
+
+    // A subscription made now, as the broadcast subscriber makes it at
+    // startup, has to work once NATS is up
+    let mut stream = coordinator.subscribe(TOPIC).await.unwrap();
+
+    // 2. NATS comes up: the node connects on its own
+    let nats = GenericImage::new("nats", "latest")
+        .with_wait_for(WaitFor::message_on_stderr("Server is ready"))
+        .with_mapped_port(nats_port, 4222.tcp())
+        .start()
+        .await
+        .expect("Failed to start NATS container");
+    wait_for_health(200, "connected").await;
+    let other_node = coordinator::backend::nats::NatsPubSub::open(config.clone())
+        .await
+        .unwrap();
+    wait_until_connected(&other_node).await;
+    round_trip(&other_node, &mut stream, b"after startup").await;
+
+    // 3. NATS goes away: the node reports it; and it comes back: the node
+    // reconnects and the same subscription carries on
+    nats.stop().await.unwrap();
+    wait_for_health(503, "disconnected").await;
+    nats.start().await.unwrap();
+    wait_for_health(200, "connected").await;
+    wait_until_connected(&other_node).await;
+    round_trip(&other_node, &mut stream, b"after reconnect").await;
+
+    drop(nats);
+    if test.is_reset() {
+        test.temp_dir.delete();
+    }
+}
+
+async fn cluster_health() -> (u16, Value) {
+    let response = reqwest::Client::builder()
+        .danger_accept_invalid_certs(true)
+        .timeout(Duration::from_secs(5))
+        .build()
+        .unwrap()
+        .get(format!("https://127.0.0.1:{HTTP_PORT}/healthz/cluster"))
+        .send()
+        .await
+        .unwrap();
+    let status = response.status().as_u16();
+    (status, response.json().await.unwrap())
+}
+
+async fn wait_for_health(status: u16, state: &str) {
+    let started = Instant::now();
+    loop {
+        let health = cluster_health().await;
+        if health == (status, json!({"coordinator": state})) {
+            return;
+        }
+        assert!(
+            started.elapsed() < Duration::from_secs(30),
+            "expected {status} {state}, still {health:?}"
+        );
+        tokio::time::sleep(Duration::from_millis(250)).await;
+    }
+}
+
+async fn wait_until_connected(coordinator: &Coordinator) {
+    let started = Instant::now();
+    while coordinator.is_connected() != Some(true) {
+        assert!(started.elapsed() < Duration::from_secs(30), "not connected");
+        tokio::time::sleep(Duration::from_millis(100)).await;
+    }
+}
+
+/// Another node publishes; this one's subscription receives it.
+async fn round_trip(from: &Coordinator, stream: &mut coordinator::PubSubStream, payload: &[u8]) {
+    from.publish(TOPIC, payload.to_vec()).await.unwrap();
+    let message = tokio::time::timeout(Duration::from_secs(10), stream.next())
+        .await
+        .expect("no message within 10 seconds")
+        .expect("subscription ended");
+    assert_eq!(message.payload(), payload);
+}

tests/src/cluster/mod.rs

@@ -2,7 +2,11 @@
  * SPDX-FileCopyrightText: 2020 Stalwart Labs LLC <[email protected]>
  *
  * SPDX-License-Identifier: AGPL-3.0-only OR LicenseRef-SEL
+ *
+ * Modified by Coffey Labs in 2026 for INBUXA.
  */
 
 pub mod broadcast;
+#[cfg(feature = "nats")]
+pub mod coordinator; // inbuxa: coordinator reconnects
 pub mod stress;

tests/src/smtp/reporting/mod.rs

@@ -2,12 +2,9 @@
  * SPDX-FileCopyrightText: 2020 Stalwart Labs LLC <[email protected]>
  *
  * SPDX-License-Identifier: AGPL-3.0-only OR LicenseRef-SEL
- *
- * Modified by Coffey Labs in 2026 for INBUXA.
  */
 
 pub mod analyze;
 pub mod dmarc;
-pub mod reschedule; // inbuxa: report reschedules and unreadable queue rows
 pub mod scheduler;
 pub mod tls;

tests/src/smtp/reporting/reschedule.rs

@@ -1,370 +0,0 @@
-/*
- * SPDX-FileCopyrightText: 2026 Coffey Labs
- *
- * SPDX-License-Identifier: AGPL-3.0-only
- */
-
-//! Rescheduling an internal DMARC or TLS report over JMAP moves its task: the
-//! task runs at the new time, x:Task/get shows the new due, and tasks due
-//! after it still run. A task queue row whose type can't be read is logged
-//! and repaired rather than stopping every task due after it, including the
-//! rows an earlier reschedule wrote with the report's object type.
-
-use crate::utils::server::{TestServer, TestServerBuilder};
-use common::{
-    Server,
-    config::smtp::report::AggregateFrequency,
-    ipc::{DmarcEvent, PolicyType, TlsEvent},
-};
-use mail_auth::{
-    common::parse::TxtRecordParser,
-    dmarc::Dmarc,
-    mta_sts::TlsRpt,
-    report::{ActionDisposition, DmarcResult, Record},
-};
-use registry::{
-    schema::{
-        enums::{TaskStoreMaintenanceType, TaskType},
-        prelude::{ObjectType, Property},
-        structs::{
-            DmarcInternalReport, DmarcReportSettings, Expression, Task, TaskStatus,
-            TaskStoreMaintenance, TlsInternalReport, TlsReportSettings,
-        },
-    },
-    types::{EnumImpl, ObjectImpl, datetime::UTCDateTime},
-};
-use serde_json::json;
-use smtp::reporting::{index::InternalReportIndex, send::MtaReportSend};
-use std::{
-    sync::Arc,
-    time::{Duration, Instant},
-};
-use store::{
-    SerializeInfallible, ValueKey,
-    write::{BatchBuilder, RegistryClass, TaskQueueClass, ValueClass, now},
-};
-use types::id::Id;
-use utils::snowflake::SnowflakeIdGenerator;
-
-#[tokio::test(flavor = "multi_thread")]
-#[serial_test::serial]
-async fn report_reschedule() {
-    let mut test = TestServerBuilder::new("smtp_report_reschedule")
-        .await
-        .with_http_listener(19057)
-        .await
-        .capture_queue()
-        .build()
-        .await;
-
-    let admin = test.account("admin");
-    admin
-        .registry_create_object(TlsReportSettings {
-            max_report_size: Expression {
-                else_: "1024".into(),
-                ..Default::default()
-            },
-            ..Default::default()
-        })
-        .await;
-    admin
-        .registry_create_object(DmarcReportSettings {
-            aggregate_max_report_size: Expression {
-                else_: "1024".into(),
-                ..Default::default()
-            },
-            ..Default::default()
-        })
-        .await;
-    admin.reload_settings().await;
-    test.reload_core();
-    test.expect_reload_settings().await;
-    let admin = test.account("admin");
-
-    // A daily DMARC and TLS report, due a day from now
-    schedule_dmarc(&test, "foobar.org").await;
-    schedule_tls(&test, "foobar.org").await;
-    let dmarc_id = wait_for_report::<DmarcInternalReport>(&test, "foobar.org").await;
-    let tls_id = wait_for_report::<TlsInternalReport>(&test, "foobar.org").await;
-
-    // Reschedule both to a few seconds from now, with a task due after them
-    let at = now() + 3;
-    let later = marker_task(&test.server, at + 3).await;
-    for (object, id, task_type) in [
-        (
-            ObjectType::DmarcInternalReport,
-            dmarc_id,
-            TaskType::DmarcReport,
-        ),
-        (ObjectType::TlsInternalReport, tls_id, TaskType::TlsReport),
-    ] {
-        admin
-            .registry_update_object(
-                object,
-                id,
-                json!({
-                    Property::DeliverAt: UTCDateTime::from_timestamp(at as i64),
-                }),
-            )
-            .await;
-
-        // x:Task/get shows the new due, and the queue row carries the task's
-        // type. Upstream wrote the report's object type there and left the
-        // task at its old due
-        let task = admin.registry_get::<Task>(id).await;
-        assert_eq!(task.object_type(), task_type);
-        assert_eq!(
-            task.due_timestamp(),
-            at,
-            "{object:?} task due not moved: {task:?}"
-        );
-        assert_eq!(
-            queue_row(&test.server, id.id(), at).await,
-            Some(task_type.to_id().serialize()),
-            "{object:?} queue row"
-        );
-    }
-
-    // Both reports go out at the new time, and the later task still runs
-    wait_until_run(&test.server, &[dmarc_id.id(), tls_id.id(), later]).await;
-    assert!(now() >= at, "the reports went out before their new time");
-    assert!(
-        admin
-            .registry_get_all::<DmarcInternalReport>()
-            .await
-            .is_empty()
-    );
-    assert!(
-        admin
-            .registry_get_all::<TlsInternalReport>()
-            .await
-            .is_empty()
-    );
-
-    // Rows an earlier reschedule may have left in a store: one with the
-    // report's object type and the task left at its old due, and one that
-    // is unreadable and has no task behind it. Neither may hold back a task
-    // due after them.
-    schedule_dmarc(&test, "foobar.net").await;
-    let dmarc_id = wait_for_report::<DmarcInternalReport>(&test, "foobar.net").await;
-    let at = now() + 2;
-    let old_due = old_style_reschedule(&test.server, dmarc_id.id(), at).await;
-    let orphan = SnowflakeIdGenerator::global_id().unwrap();
-    let mut batch = BatchBuilder::new();
-    batch.set(
-        ValueClass::TaskQueue(TaskQueueClass::Due {
-            id: orphan,
-            due: at,
-        }),
-        vec![0xff, 0xff],
-    );
-    test.server.store().write(batch.build_all()).await.unwrap();
-    let later = marker_task(&test.server, at + 2).await;
-
-    wait_until_run(&test.server, &[dmarc_id.id(), later]).await;
-    assert!(
-        admin
-            .registry_get_all::<DmarcInternalReport>()
-            .await
-            .is_empty()
-    );
-    assert_eq!(queue_row(&test.server, orphan, at).await, None);
-    assert_eq!(queue_row(&test.server, dmarc_id.id(), at).await, None);
-    assert_eq!(queue_row(&test.server, dmarc_id.id(), old_due).await, None);
-
-    // x:Task/query by type skips an unreadable row rather than failing
-    let mut batch = BatchBuilder::new();
-    let due = now() + 3600;
-    batch.set(
-        ValueClass::TaskQueue(TaskQueueClass::Due { id: orphan, due }),
-        vec![0xff, 0xff],
-    );
-    test.server.store().write(batch.build_all()).await.unwrap();
-    admin
-        .registry_query_ids(
-            ObjectType::Task,
-            vec![(Property::Type, TaskType::DmarcReport.as_str())],
-            Vec::<&str>::new(),
-        )
-        .await;
-    let mut batch = BatchBuilder::new();
-    batch.clear(ValueClass::TaskQueue(TaskQueueClass::Due {
-        id: orphan,
-        due,
-    }));
-    test.server.store().write(batch.build_all()).await.unwrap();
-
-    if test.is_reset() {
-        test.temp_dir.delete();
-    }
-}
-
-async fn schedule_dmarc(test: &TestServer, domain: &str) {
-    test.server
-        .schedule_report(DmarcEvent {
-            domain: domain.to_string(),
-            report_record: Record::new()
-                .with_source_ip("192.168.1.2".parse().unwrap())
-                .with_action_disposition(ActionDisposition::Pass)
-                .with_dmarc_dkim_result(DmarcResult::Pass)
-                .with_dmarc_spf_result(DmarcResult::Fail)
-                .with_envelope_from("[email protected]")
-                .with_envelope_to("[email protected]")
-                .with_header_from("[email protected]"),
-            dmarc_record: Arc::new(
-                Dmarc::parse(format!("v=DMARC1; p=reject; rua=mailto:reports@{domain}").as_bytes())
-                    .unwrap(),
-            ),
-            interval: AggregateFrequency::Daily,
-            span_id: 0,
-        })
-        .await;
-}
-
-async fn schedule_tls(test: &TestServer, domain: &str) {
-    test.server
-        .schedule_report(TlsEvent {
-            domain: domain.to_string(),
-            policy: PolicyType::None,
-            failure: None,
-            tls_record: Arc::new(
-                TlsRpt::parse(format!("v=TLSRPTv1;rua=mailto:reports@{domain}").as_bytes())
-                    .unwrap(),
-            ),
-            interval: AggregateFrequency::Daily,
-            span_id: 0,
-        })
-        .await;
-}
-
-trait ReportDomain: ObjectImpl {
-    fn report_domain(&self) -> &str;
-}
-
-impl ReportDomain for DmarcInternalReport {
-    fn report_domain(&self) -> &str {
-        &self.domain
-    }
-}
-
-impl ReportDomain for TlsInternalReport {
-    fn report_domain(&self) -> &str {
-        &self.domain
-    }
-}
-
-async fn wait_for_report<T: ReportDomain>(test: &TestServer, domain: &str) -> Id {
-    let admin = test.account("admin");
-    for _ in 0..100 {
-        if let Some((id, _)) = admin
-            .registry_get_all::<T>()
-            .await
-            .into_iter()
-            .find(|(_, report)| report.report_domain() == domain)
-        {
-            return id;
-        }
-        tokio::time::sleep(Duration::from_millis(100)).await;
-    }
-    panic!("No {} for {domain}", T::OBJECT.as_str());
-}
-
-/// A task that succeeds when it runs, due at `due`.
-async fn marker_task(server: &Server, due: u64) -> u64 {
-    let id = SnowflakeIdGenerator::global_id().unwrap();
-    let mut batch = BatchBuilder::new();
-    batch.schedule_task_with_id(
-        id,
-        Task::StoreMaintenance(TaskStoreMaintenance {
-            maintenance_type: TaskStoreMaintenanceType::RemoveLockDav,
-            shard_index: Some(0),
-            status: TaskStatus::at(due as i64),
-        }),
-    );
-    server.store().write(batch.build_all()).await.unwrap();
-    server.notify_task_queue();
-    id
-}
-
-/// What the reschedule before this fix wrote: the report's object type in
-/// the new queue row, and the task row left at its old due. Returns that
-/// old due.
-async fn old_style_reschedule(server: &Server, item_id: u64, at: u64) -> u64 {
-    let object_id = ObjectType::DmarcInternalReport.to_id();
-    let key = ValueClass::Registry(RegistryClass::Item { object_id, item_id });
-    let mut report = server
-        .store()
-        .get_value::<DmarcInternalReport>(ValueKey::from(key.clone()))
-        .await
-        .unwrap()
-        .unwrap();
-    let old_due = report.deliver_at().timestamp() as u64;
-    report.set_deliver_at(UTCDateTime::from_timestamp(at as i64));
-    let mut batch = BatchBuilder::new();
-    batch
-        .clear(ValueClass::TaskQueue(TaskQueueClass::Due {
-            id: item_id,
-            due: old_due,
-        }))
-        .set(
-            ValueClass::TaskQueue(TaskQueueClass::Due {
-                id: item_id,
-                due: at,
-            }),
-            object_id.serialize(),
-        )
-        .set(key, report.to_pickled_vec());
-    server.store().write(batch.build_all()).await.unwrap();
-    server.notify_task_queue();
-    old_due
-}
-
-struct RawValue(Vec<u8>);
-
-impl store::Deserialize for RawValue {
-    fn deserialize(bytes: &[u8]) -> trc::Result<Self> {
-        Ok(RawValue(bytes.to_vec()))
-    }
-}
-
-async fn queue_row(server: &Server, id: u64, due: u64) -> Option<Vec<u8>> {
-    server
-        .store()
-        .get_value::<RawValue>(ValueKey::from(ValueClass::TaskQueue(TaskQueueClass::Due {
-            id,
-            due,
-        })))
-        .await
-        .unwrap()
-        .map(|raw| raw.0)
-}
-
-async fn task_exists(server: &Server, id: u64) -> bool {
-    server
-        .store()
-        .get_value::<Task>(ValueKey::from(ValueClass::TaskQueue(
-            TaskQueueClass::Task { id },
-        )))
-        .await
-        .unwrap()
-        .is_some()
-}
-
-async fn wait_until_run(server: &Server, ids: &[u64]) {
-    let started = Instant::now();
-    loop {
-        let mut pending = Vec::new();
-        for id in ids {
-            if task_exists(server, *id).await {
-                pending.push(*id);
-            }
-        }
-        if pending.is_empty() {
-            return;
-        }
-        if started.elapsed() > Duration::from_secs(30) {
-            panic!("tasks {pending:?} never ran");
-        }
-        tokio::time::sleep(Duration::from_millis(200)).await;
-    }
-}

tests/src/store/import_export.rs

@@ -2,6 +2,8 @@
  * SPDX-FileCopyrightText: 2020 Stalwart Labs LLC <[email protected]>
  *
  * SPDX-License-Identifier: AGPL-3.0-only OR LicenseRef-SEL
+ *
+ * Modified by Coffey Labs in 2026 for INBUXA.
  */
 
 use crate::utils::{
@@ -9,14 +11,22 @@ use crate::utils::{
     server::TestServer,
     temp_dir::TempDir,
 };
-use ::registry::schema::enums::CompressionAlgo;
+use ::registry::schema::{
+    enums::{CompressionAlgo, TaskStoreMaintenanceType},
+    prelude::ObjectType,
+    structs::Task,
+};
 use ahash::AHashSet;
-use common::{DATABASE_SCHEMA_VERSION, manager::backup::BackupParams};
+use common::{
+    DATABASE_SCHEMA_VERSION,
+    manager::{SPAM_CLASSIFIER_KEY, SPAM_TRAINER_KEY, backup::BackupParams},
+};
 use store::{
     rand,
     write::{
         AnyClass, AnyKey, BatchBuilder, BlobLink, BlobOp, Operation, QueueClass, QueueEvent,
-        RegistryClass, ValueClass, key::KeySerializer,
+        RegistryClass, TaskQueueClass, ValueClass,
+        key::{DeserializeBigEndian, KeySerializer},
     },
     *,
 };
@@ -167,6 +177,50 @@ pub async fn test(test: &TestServer) {
     }
     db.write(batch.build_all()).await.unwrap();
 
+    // inbuxa: registry objects kept outside the registry subspace (archived
+    // items for undelete, spam training samples, directory entries) and the
+    // fork's own subspace. Exports used to leave the first two behind.
+    println!("Creating archived items, spam samples and fork data...");
+    let mut batch = BatchBuilder::new();
+    for item_id in [1u64, 2, 3] {
+        for object in [
+            ObjectType::ArchivedItem,
+            ObjectType::SpamTrainingSample,
+            ObjectType::Account,
+        ] {
+            batch.set(
+                ValueClass::Registry(RegistryClass::Item {
+                    object_id: object as u16,
+                    item_id,
+                }),
+                random_bytes(item_id as usize * 64),
+            );
+        }
+        batch.set(
+            ValueClass::Any(AnyClass {
+                subspace: SUBSPACE_INBUXA,
+                key: [b'U', b'x']
+                    .into_iter()
+                    .chain(item_id.to_be_bytes())
+                    .collect(),
+            }),
+            random_bytes(32),
+        );
+    }
+    db.write(batch.build_all()).await.unwrap();
+
+    // inbuxa: the trained spam classifier lives in blobs with fixed names
+    let mut named_blobs = Vec::new();
+    for key in [SPAM_CLASSIFIER_KEY, SPAM_TRAINER_KEY] {
+        let data = random_bytes(4096);
+        test.server
+            .blob_store()
+            .put_blob(key, &data, CompressionAlgo::Lz4)
+            .await
+            .unwrap();
+        named_blobs.push((key, data));
+    }
+
     // Create directory data
     println!("Creating directory data...");
     let mut batch = BatchBuilder::new();
@@ -185,6 +239,17 @@ pub async fn test(test: &TestServer) {
     println!("Calculating store hash...");
     let snapshot = Snapshot::new(&db).await;
     assert!(!snapshot.keys.is_empty(), "Store hash counts are empty",);
+    for subspace in [
+        SUBSPACE_DELETED_ITEMS,
+        SUBSPACE_SPAM_SAMPLES,
+        SUBSPACE_INBUXA,
+    ] {
+        assert!(
+            snapshot.keys.iter().any(|k| k.subspace == subspace),
+            "No test data in subspace {}",
+            char::from(subspace)
+        );
+    }
 
     // Export store
     println!("Exporting store...");
@@ -210,22 +275,188 @@ pub async fn test(test: &TestServer) {
             .finalize(),
     );
     db.write(batch.build_all()).await.unwrap();
-    test.server.core.restore(temp_dir.path.clone()).await;
+    for (key, _) in &named_blobs {
+        test.server.blob_store().delete_blob(key).await.unwrap();
+    }
+    let imported = test.server.core.restore(temp_dir.path.clone()).await;
     let mut batch = BatchBuilder::new();
     batch.clear(ValueClass::NodeId(0));
     db.write(batch.build_all()).await.unwrap();
+    for subspace in [
+        SUBSPACE_DELETED_ITEMS,
+        SUBSPACE_SPAM_SAMPLES,
+        SUBSPACE_INBUXA,
+    ] {
+        assert!(
+            imported.contains(&subspace),
+            "Subspace {} was not exported",
+            char::from(subspace)
+        );
+    }
 
     // Verify hash
     print!("Verifying store hash...");
     snapshot.assert_is_eq(&Snapshot::new(&db).await);
+    assert_named_blobs(test.server.blob_store(), &named_blobs).await;
     println!(" GREAT SUCCESS!");
 
+    // inbuxa: import the same export into a fresh store of another backend,
+    // the way a move from one database to another does it
+    #[cfg(all(feature = "rocks", feature = "sqlite"))]
+    cross_backend(test, &db, &temp_dir, &named_blobs).await;
+
     // Destroy store
+    for (key, _) in &named_blobs {
+        test.server.blob_store().delete_blob(key).await.unwrap();
+    }
     store_destroy(&db).await;
     store_assert_is_empty(&db, db.clone().into(), true).await;
     temp_dir.delete();
 }
 
+#[cfg(all(feature = "rocks", feature = "sqlite"))]
+async fn cross_backend(
+    test: &TestServer,
+    source: &Store,
+    export: &TempDir,
+    named_blobs: &[(&[u8], Vec<u8>)],
+) {
+    let source_type = std::env::var("STORE").unwrap();
+    let target_type = if source_type.eq_ignore_ascii_case("sqlite") {
+        "RocksDb"
+    } else {
+        "Sqlite"
+    };
+    println!("Importing the export into a fresh {target_type} store...");
+
+    let target_dir = TempDir::new("art_vandelay_cross_backend", true);
+    let target = Store::build(
+        crate::utils::storage::build_data_store(target_type, &target_dir.path.to_string_lossy())
+            .await,
+    )
+    .await
+    .unwrap();
+    target.create_tables().await.unwrap();
+    store_destroy(&target).await;
+
+    let mut core = test.server.core.as_ref().clone();
+    core.storage.data = target.clone();
+    core.storage.blob = target.clone().into();
+    let imported = core.restore(export.path.clone()).await;
+
+    // Counters are stored differently by the SQL and key-value backends, so
+    // compare their keys here and their values through the counter API.
+    print!("Verifying {target_type} store hash...");
+    Snapshot::new_portable(source)
+        .await
+        .assert_is_eq(&Snapshot::new_portable(&target).await);
+    for subspace in [SUBSPACE_COUNTER, SUBSPACE_QUOTA] {
+        let mut keys = Vec::new();
+        source
+            .iterate(
+                IterateParams::new(
+                    AnyKey {
+                        subspace,
+                        key: vec![0u8],
+                    },
+                    AnyKey {
+                        subspace,
+                        key: vec![u8::MAX; 10],
+                    },
+                )
+                .no_values(),
+                |key, _| {
+                    keys.push(key.to_vec());
+                    Ok(true)
+                },
+            )
+            .await
+            .unwrap();
+        for key in keys {
+            let class = || {
+                ValueClass::Any(AnyClass {
+                    subspace,
+                    key: key.clone(),
+                })
+            };
+            assert_eq!(
+                source.get_counter(class()).await.unwrap(),
+                target.get_counter(class()).await.unwrap(),
+                "Counter mismatch in {} for {key:?}",
+                char::from(subspace)
+            );
+        }
+    }
+    assert_named_blobs(&core.storage.blob, named_blobs).await;
+    println!(" GREAT SUCCESS!");
+
+    // The search index isn't exported; the import queues its rebuild
+    let queued = core.queue_reindex(&imported).await;
+    let expected = [
+        TaskStoreMaintenanceType::ReindexAccounts,
+        TaskStoreMaintenanceType::ReindexTelemetry,
+    ];
+    assert_eq!(queued, expected);
+    let mut task_ids = Vec::new();
+    target
+        .iterate(
+            IterateParams::new(
+                AnyKey {
+                    subspace: SUBSPACE_TASK_QUEUE,
+                    key: vec![0u8],
+                },
+                AnyKey {
+                    subspace: SUBSPACE_TASK_QUEUE,
+                    key: vec![u8::MAX; 20],
+                },
+            )
+            .no_values(),
+            |key, _| {
+                if key.deserialize_be_u64(0)? == 0 {
+                    task_ids.push(key.deserialize_be_u64(U64_LEN)?);
+                }
+                Ok(true)
+            },
+        )
+        .await
+        .unwrap();
+    let mut found = Vec::new();
+    for id in task_ids {
+        match target
+            .get_value::<Task>(ValueKey::from(ValueClass::TaskQueue(
+                TaskQueueClass::Task { id },
+            )))
+            .await
+            .unwrap()
+        {
+            Some(Task::StoreMaintenance(task)) => found.push(task.maintenance_type),
+            other => panic!("Unexpected task {other:?}"),
+        }
+    }
+    found.sort_by_key(|t| *t as u16);
+    assert_eq!(found, expected, "Queued tasks don't match");
+
+    store_destroy(&target).await;
+    drop(core);
+    drop(target);
+    target_dir.delete();
+}
+
+async fn assert_named_blobs(blob_store: &BlobStore, named_blobs: &[(&[u8], Vec<u8>)]) {
+    for (key, data) in named_blobs {
+        assert_eq!(
+            blob_store
+                .get_blob(key, 0..usize::MAX)
+                .await
+                .unwrap()
+                .as_ref(),
+            Some(data),
+            "Blob {} was not restored",
+            String::from_utf8_lossy(key)
+        );
+    }
+}
+
 #[derive(Debug, PartialEq, Eq)]
 struct Snapshot {
     keys: AHashSet<KeyValue>,
@@ -240,7 +471,19 @@ struct KeyValue {
 
 impl Snapshot {
     async fn new(db: &Store) -> Self {
-        let is_sql = db.is_sql();
+        Self::build(db, !db.is_sql(), true).await
+    }
+
+    /// Comparable across backends: no counter values, which the SQL and
+    /// key-value stores encode differently, and no blobs, which only live in
+    /// the data store when it doubles as the blob store.
+    #[cfg(all(feature = "rocks", feature = "sqlite"))]
+    async fn new_portable(db: &Store) -> Self {
+        Self::build(db, false, false).await
+    }
+
+    async fn build(db: &Store, counter_values: bool, with_blobs: bool) -> Self {
+        let is_sql = !counter_values;
 
         let mut keys = AHashSet::new();
 
@@ -265,7 +508,12 @@ impl Snapshot {
             (SUBSPACE_QUOTA, !is_sql),
             (SUBSPACE_REPORT_OUT, true),
             (SUBSPACE_REPORT_IN, true),
+            (SUBSPACE_DIRECTORY, true),
+            (SUBSPACE_INBUXA, true),
         ] {
+            if subspace == SUBSPACE_BLOBS && !with_blobs {
+                continue;
+            }
             let from_key = AnyKey {
                 subspace,
                 key: vec![0u8],

tests/src/store/query.rs

@@ -2,6 +2,8 @@
  * SPDX-FileCopyrightText: 2020 Stalwart Labs LLC <[email protected]>
  *
  * SPDX-License-Identifier: AGPL-3.0-only OR LicenseRef-SEL
+ *
+ * Modified by Coffey Labs in 2026 for INBUXA.
  */
 
 use crate::{store::deflate_test_resource, utils::server::TestServer};
@@ -122,6 +124,10 @@ pub async fn test(test: &TestServer) {
     println!("Running global id filtering tests...");
     test_global(store.clone()).await;
 
+    // inbuxa: trace documents as the index task builds them
+    println!("Running trace document tests...");
+    test_trace_documents(store.clone()).await;
+
     // Large document insert test
     println!("Running large document insert tests...");
     let mut large_text = String::with_capacity(20 * 1024 * 1024);
@@ -809,3 +815,160 @@ async fn test_global(store: SearchStore) {
         AHashSet::from_iter([3, 4, 5])
     );
 }
+
+// inbuxa: MON-16: documents built by the index task from stored traces go
+// into every search backend (the SQL backends type etyp and qid as BIGINT)
+// and are found again by queue id and keyword.
+async fn test_trace_documents(store: SearchStore) {
+    use registry::schema::{
+        enums::SearchTracingField,
+        structs::{
+            Trace, TraceEvent, TraceKeyValue, TraceValue, TraceValueString,
+            TraceValueUnsignedInt,
+        },
+    };
+    use services::task_manager::index::trace_search_document;
+    use trc::{DeliveryEvent, EventType, Key, SmtpEvent};
+
+    let kv_u = |key: Key, value: u64| TraceKeyValue {
+        key,
+        value: TraceValue::UnsignedInt(TraceValueUnsignedInt { value }),
+    };
+    let kv_s = |key: Key, value: &str| TraceKeyValue {
+        key,
+        value: TraceValue::String(TraceValueString {
+            value: value.to_string(),
+        }),
+    };
+    let event = |event: EventType, key_values: Vec<TraceKeyValue>| TraceEvent {
+        event,
+        key_values: key_values.into(),
+        ..Default::default()
+    };
+    let fields = [
+        SearchTracingField::EventType,
+        SearchTracingField::QueueId,
+        SearchTracingField::Keywords,
+    ];
+
+    // An SMTP session that queued two messages, and a delivery attempt
+    let session = Trace {
+        events: vec![
+            event(
+                EventType::Smtp(SmtpEvent::ConnectionStart),
+                vec![kv_s(Key::RemoteIp, "192.0.2.7")],
+            ),
+            event(
+                EventType::Smtp(SmtpEvent::MailFrom),
+                vec![kv_s(Key::From, "[email protected]")],
+            ),
+            event(
+                EventType::Smtp(SmtpEvent::RcptTo),
+                vec![kv_u(Key::QueueId, 9_000_000_001), kv_s(Key::To, "[email protected]")],
+            ),
+            event(
+                EventType::Smtp(SmtpEvent::RcptTo),
+                vec![kv_u(Key::QueueId, 9_000_000_002)],
+            ),
+        ]
+        .into(),
+    };
+    let delivery = Trace {
+        events: vec![event(
+            EventType::Delivery(DeliveryEvent::AttemptStart),
+            vec![kv_u(Key::QueueId, 9_000_000_003), kv_s(Key::Hostname, "relay.example.net")],
+        )]
+        .into(),
+    };
+    let documents = vec![
+        trace_search_document(100, &session, &fields),
+        trace_search_document(101, &delivery, &fields),
+    ];
+    assert!(
+        documents
+            .iter()
+            .all(|d| d.has_field(&SearchField::Tracing(TracingSearchField::QueueId))
+                && d.has_field(&SearchField::Tracing(TracingSearchField::EventType))),
+        "trace documents carry a queue id and an event type"
+    );
+    store.index(documents).await.unwrap();
+    if let SearchStore::ElasticSearch(store) = &store {
+        store.refresh_index(SearchIndex::Tracing).await.unwrap();
+    }
+
+    let query = |filters: Vec<SearchFilter>| {
+        let store = store.clone();
+        async move {
+            store
+                .query_global(
+                    SearchQuery::new(SearchIndex::Tracing)
+                        .with_filter(SearchFilter::ge(SearchField::Id, 100u64))
+                        .with_filters(filters),
+                )
+                .await
+                .unwrap()
+                .into_iter()
+                .collect::<AHashSet<_>>()
+        }
+    };
+    // By queue id, the way x:Trace/query asks: the queue id column, or any
+    // queue id in the keywords
+    let by_queue_id = |queue_id: u64| {
+        vec![
+            SearchFilter::Or,
+            SearchFilter::eq(TracingSearchField::QueueId, queue_id),
+            SearchFilter::has_text(
+                TracingSearchField::Keywords,
+                queue_id.to_string(),
+                Language::None,
+            ),
+            SearchFilter::End,
+        ]
+    };
+    assert_eq!(query(by_queue_id(9_000_000_001)).await, AHashSet::from_iter([100]));
+    assert_eq!(query(by_queue_id(9_000_000_002)).await, AHashSet::from_iter([100]));
+    assert_eq!(query(by_queue_id(9_000_000_003)).await, AHashSet::from_iter([101]));
+    assert_eq!(query(by_queue_id(9_000_000_004)).await, AHashSet::new());
+    assert_eq!(
+        query(vec![SearchFilter::eq(TracingSearchField::QueueId, 9_000_000_003u64)]).await,
+        AHashSet::from_iter([101])
+    );
+    // By opening event type
+    assert_eq!(
+        query(vec![SearchFilter::eq(
+            TracingSearchField::EventType,
+            EventType::Delivery(DeliveryEvent::AttemptStart).to_id() as u64,
+        )])
+        .await,
+        AHashSet::from_iter([101])
+    );
+    // By keyword: an address, lowercased, and its domain
+    assert_eq!(
+        query(vec![SearchFilter::has_text(
+            TracingSearchField::Keywords,
+            "example.org",
+            Language::None,
+        )])
+        .await,
+        AHashSet::from_iter([100])
+    );
+    assert_eq!(
+        query(vec![SearchFilter::has_text(
+            TracingSearchField::Keywords,
+            "relay.example.net",
+            Language::None,
+        )])
+        .await,
+        AHashSet::from_iter([101])
+    );
+
+    for id in [100u64, 101] {
+        store
+            .unindex(
+                SearchQuery::new(SearchIndex::Tracing)
+                    .with_filter(SearchFilter::eq(SearchField::Id, id)),
+            )
+            .await
+            .unwrap();
+    }
+}

tests/src/system/monitoring.rs

@@ -148,6 +148,73 @@ pub async fn test(test: &mut TestServer) {
         "test 9: to"
     );
 
+    // MON-16: the queueId filter finds the traces that name a queue id (the
+    // session that queued the message and its delivery attempt) through the
+    // search index, given as a string or a number (the index column is an
+    // integer)
+    fn queue_ids(value: &Value, out: &mut Vec<u64>) {
+        match value {
+            Value::Object(map) => {
+                if map.get("key").and_then(|k| k.as_str()) == Some("queueId")
+                    && let Some(id) = map
+                        .get("value")
+                        .and_then(|v| v.get("value").unwrap_or(v).as_u64())
+                {
+                    out.push(id);
+                }
+                map.values().for_each(|v| queue_ids(v, out));
+            }
+            Value::Array(list) => list.iter().for_each(|v| queue_ids(v, out)),
+            _ => {}
+        }
+    }
+    let with_ids = traces
+        .iter()
+        .map(|t| {
+            let mut ids = Vec::new();
+            queue_ids(t, &mut ids);
+            (t["id"].as_str().unwrap().to_string(), ids)
+        })
+        .collect::<Vec<_>>();
+    let queue_id = with_ids
+        .iter()
+        .find_map(|(_, ids)| ids.first().copied())
+        .expect("MON-16: a trace with a queue id");
+    let mut expected = with_ids
+        .iter()
+        .filter(|(_, ids)| ids.contains(&queue_id))
+        .map(|(id, _)| id.clone())
+        .collect::<Vec<_>>();
+    expected.sort();
+    for filter in [json!(queue_id.to_string()), json!(queue_id)] {
+        let response = admin
+            .jmap_method_call("x:Trace/query", json!({"filter": {"queueId": filter}}))
+            .await;
+        let mut found = response
+            .0
+            .pointer("/methodResponses/0/1/ids")
+            .and_then(|ids| ids.as_array())
+            .map(|ids| {
+                ids.iter()
+                    .filter_map(|id| id.as_str().map(str::to_string))
+                    .collect::<Vec<_>>()
+            })
+            .unwrap_or_default();
+        found.sort();
+        assert_eq!(found, expected, "MON-16: queueId {filter}: {response:?}");
+    }
+    let response = admin
+        .jmap_method_call(
+            "x:Trace/query",
+            json!({"filter": {"queueId": (queue_id ^ 0x5a5a_5a5a).to_string()}}),
+        )
+        .await;
+    assert_eq!(
+        response.0.pointer("/methodResponses/0/1/ids"),
+        Some(&json!([])),
+        "MON-16: an unknown queue id"
+    );
+
     // Acceptance test 24: destroy removes a trace; create is refused
     let trace_id = traces[0]["id"].as_str().unwrap().to_string();
     let response = admin