Author SHA1 Message Date
jcoffey-dev f398d95062 Merge pull request 'Hotfix 2026.9.24.4: report reschedules keep the task queue readable' (#49) from hotfix/2026.9.24.4 into release/2026.9.24.4
publish / version (push) Successful in 41s
publish / publish (push) Successful in 56m8s
publish / release (push) Successful in 6s
publish / binaries (push) Successful in 49s
2026-09-25 01:43:57 +00:00
jcoffey-dev 81ec917d74 Mark task_manager/manager.rs as modified for the AGPL notice
ci / fork-checks (pull_request) Successful in 22s
ci / build (pull_request) Successful in 3m37s
2026-09-24 18:39:42 -07:00
jcoffey-dev e2ab26ad19 Release 2026.9.24.4
ci / fork-checks (pull_request) Failing after 29s
ci / build (pull_request) Canceled after 3m15s
2026-09-24 18:35:50 -07:00
jcoffey-dev 0b4aa9c084 Report reschedules keep the task queue readable
Setting deliverAt on an internal DMARC or TLS report wrote the new task
queue row with the report's object type (0x21, 0x6e) instead of the task
type (7, 8), and left the task row at its old due. The task manager's scan
failed on that row with store.data-corruption ("Failed to iterate over task
queue"), and because the error ended the whole scan, every task due after
the row stopped running on every node.

- reschedule_ops writes the new queue row through schedule_task_with_id, so
  it carries the task type and the task row gets the new due. It removes
  the row the task is actually queued under (the task's due, which differs
  from deliverAt once the task has been retried) and any row an earlier
  reschedule left at deliverAt.
- x:DmarcInternalReport/set and x:TlsInternalReport/set lock the report's
  task while they move it, as x:Task/set does, refuse while the report is
  being sent, release the locks however the request ends, and wake the task
  manager.
- The task manager logs a queue row it can't read (id, due, key, value) and
  skips it instead of ending the scan. It then repairs the row from its task:
  the row is rewritten with the task's type, and a row with no task behind
  it is removed. A row holding a report's object type for a report task is
  what the old reschedule wrote: the task is moved to that row's time, as
  the reschedule intended, and its old queue row is removed. Stores that
  already hold such a row recover on their own once it comes due.
- x:Task/query with a type filter skips an unreadable row instead of
  failing.

Test: smtp::reporting::reschedule (RocksDB and PostgreSQL). It fails on
main: x:Task/get shows the old due, and with that check removed, neither
report nor a later task ever runs.

(cherry picked from commit 1a7859a8cc)
2026-09-24 18:35:50 -07:00
jcoffey-dev 4e6c8b916e Publish: accept tags on release/* branches for hotfix releases
The publish workflow only built a tag whose commit is on main. That keeps
every image tied to reviewed code, but it means production can only get a
fix together with everything that has landed on main since its release.

A tag on a release/* branch is now accepted too. A hotfix branch starts at
an earlier release tag, takes fixes through pull requests into it (so the
code is still reviewed and CI-tested before it is tagged), bumps
brand_version! and is tagged there. The tag must still equal
v<brand_version!>, and the step prints which branch it was found on.

A tag runs the workflow file from its own commit, so a hotfix branch that
starts before this change needs this commit cherry-picked onto it before
its tag is pushed.

(cherry picked from commit 4b85113262)
2026-09-24 18:31:52 -07:00
18 changed files with 690 additions and 816 deletions
+27 -73
View File
@@ -3,36 +3,22 @@
# whether a person pushed it or weekly-release.yml created it through the # whether a person pushed it or weekly-release.yml created it through the
# releases API. # releases API.
# #
# The image is multi-arch (linux/amd64, linux/arm64), built by two jobs on # The image is multi-arch (linux/amd64, linux/arm64) as before, but built in
# the image-build runner rather than one buildx run for both. The Dockerfile's # one buildx run on host1 instead of one native runner per architecture: the
# builder stage runs on the build platform and cross-compiles with an aarch64 # Dockerfile's builder stage runs on the build platform and cross-compiles
# linker, so only the small final stage (apt, setcap) goes through QEMU for # with an aarch64 linker, so only the small final stage (apt, setcap) goes
# arm64 -- but two release builds (LTO, one codegen unit) side by side on one # through QEMU for arm64. No digest-joining job is needed.
# machine each take twice as long. Production runs amd64, so amd64 goes first
# and on its own:
# * publish-amd64 pushes :<version>-amd64 and :<version>, a plain amd64
# image, as soon as its build is done. A deploy can start from it.
# * publish-arm64 then builds arm64, pushes :<version>-arm64, and replaces
# :<version> with the two-platform index. :latest moves only here, so it
# never names an image without arm64.
#
# Both jobs use one BuildKit builder, `gitea-builder`, whose container
# (buildx_buildkit_gitea-builder0) and state volume stay on the runner's host
# between jobs: a job container's `buildx create` finds the existing container
# and reuses it and its cache. The dependency build (`cargo chef cook`) is
# keyed on the recipe, which only a dependency change alters, so a release
# normally compiles just the workspace. Removing that container or its volume
# costs the next release a cold build, nothing more. The planner and dependency
# layers for the build platform are shared, so arm64 also reuses what amd64
# just did where it can.
# #
# Two guards before anything is pushed: # Two guards before anything is pushed:
# * the tag must be v<brand_version!>. The version is a string in # * the tag must be v<brand_version!>. The version is a string in
# crates/types/src/branding.rs, not Cargo.toml, and the image is tagged # 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 # with it, so a tag beside an unbumped macro would publish an image that
# reports a different version from its tag. # reports a different version from its tag.
# * the tag must be on main, so an image never describes code that was never # * the tag must be on main or on a release/* branch, so an image never
# reviewed onto the default branch. # describes code that was never reviewed onto one of them. A release/*
# 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 # :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. # (or by hand for a real release); there are no prerelease tags here.
@@ -74,12 +60,17 @@ jobs:
echo "Refusing to publish an image that would report the wrong version." >&2 echo "Refusing to publish an image that would report the wrong version." >&2
exit 1 exit 1
fi fi
git merge-base --is-ancestor "$(git rev-parse "${TAG}^{commit}")" origin/main \ commit="$(git rev-parse "${TAG}^{commit}")"
|| { echo "$TAG is not on main" >&2; exit 1; } on=""
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" >> "$GITHUB_OUTPUT"
echo "version $V" echo "version $V"
publish-amd64: publish:
needs: [version] needs: [version]
runs-on: docker runs-on: docker
container: container:
@@ -98,15 +89,16 @@ jobs:
test -n "$REGISTRY" && test -n "$VERSION" test -n "$REGISTRY" && test -n "$VERSION"
test -n "$PACKAGE_TOKEN" || { echo "PACKAGE_TOKEN secret is not set on this repository" >&2; exit 1; } test -n "$PACKAGE_TOKEN" || { echo "PACKAGE_TOKEN secret is not set on this repository" >&2; exit 1; }
echo "$PACKAGE_TOKEN" | docker login -u jcoffey-dev --password-stdin "$REGISTRY" echo "$PACKAGE_TOKEN" | docker login -u jcoffey-dev --password-stdin "$REGISTRY"
docker run --privileged --rm tonistiigi/binfmt --install arm64
docker buildx create --use --name gitea-builder --driver docker-container || docker buildx use gitea-builder docker buildx create --use --name gitea-builder --driver docker-container || docker buildx use gitea-builder
# Attestations off, as before: they add manifests of their own, and the # Attestations off, as before: they add manifests of their own to the
# index should hold the two images and nothing else. # index, and the index should hold the two images and nothing else.
- run: | - run: |
docker buildx build \ docker buildx build \
--platform linux/amd64 \ --platform linux/amd64,linux/arm64 \
--provenance=false --sbom=false \ --provenance=false --sbom=false \
--tag "$IMAGE:$VERSION-amd64" \
--tag "$IMAGE:$VERSION" \ --tag "$IMAGE:$VERSION" \
--tag "$IMAGE:latest" \
--push . --push .
docker buildx imagetools inspect "$IMAGE:$VERSION" docker buildx imagetools inspect "$IMAGE:$VERSION"
# Gitea keeps a container package on its owner; linking it shows it on # Gitea keeps a container package on its owner; linking it shows it on
@@ -119,47 +111,11 @@ jobs:
- if: always() - if: always()
run: docker logout "$REGISTRY" || true run: docker logout "$REGISTRY" || true
publish-arm64:
needs: [version, publish-amd64]
runs-on: docker
container:
image: docker:28-cli@sha256:625d9431a9f54c5a2bc90f24f0e1c3d55b1349fd857dd85035f98c2c9acbdd4d # 28-cli
volumes:
- /var/run/docker.sock:/var/run/docker.sock
env:
DOCKER_BUILDKIT: "1"
REGISTRY: ${{ vars.REGISTRY }}
IMAGE: ${{ vars.REGISTRY }}/${{ github.repository }}
VERSION: ${{ needs.version.outputs.version }}
PACKAGE_TOKEN: ${{ secrets.PACKAGE_TOKEN }}
steps:
- uses: coffey-labs/actions/checkout@fab0c4d45e0162963965f1555df27b7bed5e20ec
- run: |
echo "$PACKAGE_TOKEN" | docker login -u jcoffey-dev --password-stdin "$REGISTRY"
docker run --privileged --rm tonistiigi/binfmt --install arm64
docker buildx create --use --name gitea-builder --driver docker-container || docker buildx use gitea-builder
# The index is built from the two per-architecture tags rather than from
# :<version>, which by now is the amd64 image and would be read as such.
- run: |
docker buildx build \
--platform linux/arm64 \
--provenance=false --sbom=false \
--tag "$IMAGE:$VERSION-arm64" \
--push .
docker buildx imagetools create \
--tag "$IMAGE:$VERSION" \
--tag "$IMAGE:latest" \
"$IMAGE:$VERSION-amd64" "$IMAGE:$VERSION-arm64"
docker buildx imagetools inspect "$IMAGE:$VERSION"
- if: always()
run: docker logout "$REGISTRY" || true
# The weekly release creates its Release (and so the tag) first; a tag # The weekly release creates its Release (and so the tag) first; a tag
# pushed by hand has none. Either way the tag ends up with exactly one # pushed by hand has none. Either way the tag ends up with exactly one
# Release, created once the amd64 image exists so its pull instructions # Release, created after the image exists so its pull instructions work.
# work; arm64 and the binaries follow.
release: release:
needs: [version, publish-amd64] needs: [version, publish]
runs-on: light runs-on: light
container: container:
image: python:3.13-slim@sha256:8d9d0b8bcf6506481eae4907c18f5e3e7902e629f5f6d684f9e7c32e85e3ddf0 # 3.13-slim image: python:3.13-slim@sha256:8d9d0b8bcf6506481eae4907c18f5e3e7902e629f5f6d684f9e7c32e85e3ddf0 # 3.13-slim
@@ -183,9 +139,7 @@ jobs:
except urllib.error.HTTPError as e: except urllib.error.HTTPError as e:
if e.code != 404: raise if e.code != 404: raise
image = f"{os.environ['REGISTRY']}/{os.environ['REPO']}:{version}" image = f"{os.environ['REGISTRY']}/{os.environ['REPO']}:{version}"
body = (f"Container image: `{image}` (linux/amd64, linux/arm64); also `:latest`. " body = (f"Container image: `{image}` (linux/amd64, linux/arm64); also `:latest`.\n\n"
"amd64 is published first; arm64 is added to the same tag when its build "
"finishes, and `:latest` moves then.\n\n"
"Binaries for a host install are attached: `inbuxa-linux-amd64.tar.gz` and " "Binaries for a host install are attached: `inbuxa-linux-amd64.tar.gz` and "
"`inbuxa-linux-arm64.tar.gz`, with `SHA256SUMS`. Each is the binary out of this " "`inbuxa-linux-arm64.tar.gz`, with `SHA256SUMS`. Each is the binary out of this "
"release's image for that architecture, so it is the same build. The image " "release's image for that architecture, so it is the same build. The image "
@@ -208,7 +162,7 @@ jobs:
# `docker create` does not start anything, so pulling an arm64 image on an # `docker create` does not start anything, so pulling an arm64 image on an
# amd64 runner and copying a file out of it needs no emulation. # amd64 runner and copying a file out of it needs no emulation.
binaries: binaries:
needs: [version, publish-arm64, release] needs: [version, publish, release]
runs-on: docker runs-on: docker
container: container:
image: docker:28-cli@sha256:625d9431a9f54c5a2bc90f24f0e1c3d55b1349fd857dd85035f98c2c9acbdd4d # 28-cli image: docker:28-cli@sha256:625d9431a9f54c5a2bc90f24f0e1c3d55b1349fd857dd85035f98c2c9acbdd4d # 28-cli
+6 -25
View File
@@ -23,13 +23,6 @@ use utils::{UnwrapFailure, codec::leb128::Leb128_};
pub(super) const MAGIC_MARKER: u8 = 123; 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)] #[derive(Debug, Clone, Copy, Hash, PartialEq, Eq)]
pub(super) enum Family { pub(super) enum Family {
Data = 0, Data = 0,
@@ -150,21 +143,15 @@ impl Core {
.await .await
.failed("Failed to iterate over data store"); .failed("Failed to iterate over data store");
// inbuxa: the trained spam classifier and its trainer state are for hash in blobs {
// 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 if let Some(blob) = blob_store
.get_blob(&key, 0..usize::MAX) .get_blob(hash.as_slice(), 0..usize::MAX)
.await .await
.failed("Failed to get blob") .failed("Failed to get blob")
{ {
writer.send((key, blob)).failed("Failed to send key"); writer
.send((hash.as_slice().to_vec(), blob))
.failed("Failed to send key");
} }
} }
}), }),
@@ -336,13 +323,7 @@ impl Family {
SUBSPACE_REGISTRY_IDX, SUBSPACE_REGISTRY_IDX,
SUBSPACE_REGISTRY_PK, SUBSPACE_REGISTRY_PK,
SUBSPACE_DIRECTORY, SUBSPACE_DIRECTORY,
// inbuxa: registry objects the upstream list left out, so an store::SUBSPACE_INBUXA, // inbuxa: masked email
// 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::Changelog => &[SUBSPACE_LOGS],
Family::Queue => &[SUBSPACE_QUEUE_MESSAGE, SUBSPACE_QUEUE_EVENT], Family::Queue => &[SUBSPACE_QUEUE_MESSAGE, SUBSPACE_QUEUE_EVENT],
+4 -11
View File
@@ -54,13 +54,6 @@ Options:
-o, --console Open the store console -o, --console Open the store console
-h, --help Print help -h, --help Print help
-V, --version Print version -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.
"# "#
); );
@@ -263,10 +256,10 @@ impl BootManager {
telemetry.enable(); telemetry.enable();
// Parse settings and restore // Parse settings and restore
let core = Box::pin(Core::parse(&mut bootstrap, storage)).await; Box::pin(Core::parse(&mut bootstrap, storage))
let imported = core.restore(path).await; .await
// inbuxa: the search index isn't exported; rebuild it .restore(path)
core.queue_reindex(&imported).await; .await;
std::process::exit(0); std::process::exit(0);
} }
StoreOp::Console => { StoreOp::Console => {
+10 -79
View File
@@ -9,22 +9,15 @@
use super::backup::MAGIC_MARKER; use super::backup::MAGIC_MARKER;
use crate::{Core, DATABASE_SCHEMA_VERSION}; use crate::{Core, DATABASE_SCHEMA_VERSION};
use lz4_flex::frame::FrameDecoder; use lz4_flex::frame::FrameDecoder;
use registry::{ use registry::schema::enums::CompressionAlgo;
schema::{
enums::{CompressionAlgo, TaskStoreMaintenanceType},
structs::{Task, TaskStatus, TaskStoreMaintenance},
},
types::EnumImpl,
};
use std::{ use std::{
fs::File, fs::File,
io::{BufReader, ErrorKind, Read}, io::{BufReader, ErrorKind, Read},
path::{Path, PathBuf}, path::{Path, PathBuf},
}; };
use store::{ use store::{
BlobStore, IterateParams, SUBSPACE_BLOBS, SUBSPACE_COUNTER, SUBSPACE_INDEXES, BlobStore, IterateParams, SUBSPACE_BLOBS, SUBSPACE_COUNTER, SUBSPACE_INDEXES, SUBSPACE_QUOTA,
SUBSPACE_PROPERTY, SUBSPACE_QUOTA, SUBSPACE_REGISTRY_PK, SUBSPACE_TELEMETRY_SPAN, Store, SUBSPACE_REGISTRY_PK, Store, U32_LEN,
U32_LEN,
write::{ write::{
AnyClass, AnyKey, BatchBuilder, ValueClass, AnyClass, AnyKey, BatchBuilder, ValueClass,
key::{DeserializeBigEndian, is_node_id_key}, key::{DeserializeBigEndian, is_node_id_key},
@@ -34,9 +27,7 @@ use types::{collection::Collection, field::Field};
use utils::{UnwrapFailure, failed}; use utils::{UnwrapFailure, failed};
impl Core { impl Core {
/// Imports an export into an empty store and returns the subspaces it pub async fn restore(&self, src: PathBuf) {
/// wrote. inbuxa: the caller hands them to [`Core::queue_reindex`].
pub async fn restore(&self, src: PathBuf) -> Vec<u8> {
// Backup the core // Backup the core
let paths = if src.is_dir() { let paths = if src.is_dir() {
let mut paths = Vec::new(); let mut paths = Vec::new();
@@ -73,13 +64,6 @@ impl Core {
std::process::exit(1); 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(); let mut tasks = Vec::new();
for path in paths { for path in paths {
let storage = self.storage.clone(); let storage = self.storage.clone();
@@ -92,54 +76,6 @@ impl Core {
for task in tasks { for task in tasks {
task.await.failed("Failed to wait for task"); 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
} }
} }
@@ -189,22 +125,17 @@ async fn restore_file(store: Store, blob_store: BlobStore, path: &Path) {
} }
SUBSPACE_COUNTER | SUBSPACE_QUOTA => { SUBSPACE_COUNTER | SUBSPACE_QUOTA => {
while let Some((key, value)) = reader.next() { while let Some((key, value)) = reader.next() {
let class = ValueClass::Any(AnyClass { batch.add(
ValueClass::Any(AnyClass {
subspace: reader.subspace, subspace: reader.subspace,
key, key,
}); }),
let value = u64::from_le_bytes( u64::from_le_bytes(
value value
.try_into() .try_into()
.expect("Failed to deserialize counter/quota"), .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() { if batch.is_large_batch() {
store store
.write(batch.build_all()) .write(batch.build_all())
+2 -17
View File
@@ -427,24 +427,9 @@ pub(crate) async fn trace_query(
} }
None => false, None => false,
}, },
// The queue id column is an integer on every search backend, and Property::QueueId => match value.as_str() {
// 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) => { Some(queue_id) => {
search.extend([ search.push(SearchFilter::eq(TracingSearchField::QueueId, queue_id.to_string()));
SearchFilter::Or,
SearchFilter::eq(TracingSearchField::QueueId, queue_id),
SearchFilter::has_text(
TracingSearchField::Keywords,
queue_id.to_string(),
nlp::language::Language::None,
),
SearchFilter::End,
]);
true true
} }
None => false, None => false,
+63 -5
View File
@@ -2,6 +2,8 @@
* SPDX-FileCopyrightText: 2020 Stalwart Labs LLC <[email protected]> * SPDX-FileCopyrightText: 2020 Stalwart Labs LLC <[email protected]>
* *
* SPDX-License-Identifier: AGPL-3.0-only OR LicenseRef-SEL * SPDX-License-Identifier: AGPL-3.0-only OR LicenseRef-SEL
*
* Modified by Coffey Labs in 2026 for INBUXA.
*/ */
use crate::{ use crate::{
@@ -15,22 +17,42 @@ use jmap_proto::{error::set::SetError, types::state::State};
use jmap_tools::{Key, Value}; use jmap_tools::{Key, Value};
use registry::{ use registry::{
jmap::IntoValue, jmap::IntoValue,
schema::prelude::{Object, ObjectInner, ObjectType, Property}, schema::{
prelude::{Object, ObjectInner, ObjectType, Property},
structs::Task,
},
types::{EnumImpl, datetime::UTCDateTime}, types::{EnumImpl, datetime::UTCDateTime},
}; };
use services::task_manager::lock::TaskLockManager;
use smtp::reporting::index::{ExternalReportIndex, InternalReportIndex}; use smtp::reporting::index::{ExternalReportIndex, InternalReportIndex};
use std::str::FromStr; use std::str::FromStr;
use store::{ use store::{
U64_LEN, ValueKey, U64_LEN, ValueKey,
registry::{RegistryFilter, RegistryFilterValue, RegistryQuery}, registry::{RegistryFilter, RegistryFilterValue, RegistryQuery},
write::{BatchBuilder, RegistryClass, ValueClass, key::KeySerializer}, write::{BatchBuilder, RegistryClass, TaskQueueClass, ValueClass, key::KeySerializer},
}; };
use trc::AddContext; use trc::AddContext;
use types::id::Id; use types::id::Id;
pub(crate) async fn report_set( pub(crate) async fn report_set(
mut set: RegistrySetResponse<'_>, set: RegistrySetResponse<'_>,
) -> trc::Result<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(); let object_id = set.object_type.to_id();
// Reports cannot be created // Reports cannot be created
@@ -89,12 +111,45 @@ pub(crate) async fn report_set(
.get_value::<Object>(ValueKey::from(key.clone())) .get_value::<Object>(ValueKey::from(key.clone()))
.await? .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 { match &mut report_obj.inner {
ObjectInner::DmarcInternalReport(report) => { ObjectInner::DmarcInternalReport(report) => {
report.reschedule_ops(&mut batch, item_id, report_obj.revision, deliver_at); report.reschedule_ops(
&mut batch,
item_id,
report_obj.revision,
deliver_at,
queued.as_ref(),
);
} }
ObjectInner::TlsInternalReport(report) => { ObjectInner::TlsInternalReport(report) => {
report.reschedule_ops(&mut batch, item_id, report_obj.revision, deliver_at); report.reschedule_ops(
&mut batch,
item_id,
report_obj.revision,
deliver_at,
queued.as_ref(),
);
} }
_ => {} _ => {}
} }
@@ -156,6 +211,9 @@ pub(crate) async fn report_set(
.write(batch.build_all()) .write(batch.build_all())
.await .await
.caused_by(trc::location!())?; .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) Ok(set)
+4 -9
View File
@@ -463,15 +463,10 @@ pub(crate) async fn task_query(
.set_values(typ.is_some()), .set_values(typ.is_some()),
|key, value| { |key, value| {
if let Some(typ) = typ { if let Some(typ) = typ {
let task_type = // inbuxa: a row whose type can't be read matches no type
TaskType::from_id(value.deserialize_be_u16(0)?).ok_or_else(|| { // filter; the task manager logs and repairs it
trc::StoreEvent::DataCorruption let task_type = value.deserialize_be_u16(0).ok().and_then(TaskType::from_id);
.into_err() if task_type != Some(typ) {
.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); return Ok(true);
} }
} }
+3 -24
View File
@@ -26,7 +26,7 @@ pub fn spawn_broadcast_subscriber(inner: Arc<Inner>, mut shutdown_rx: watch::Rec
}; };
tokio::spawn(async move { tokio::spawn(async move {
let mut retry_count: u32 = 0; let mut retry_count = 0;
trc::event!(Cluster(ClusterEvent::SubscriberStart)); 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( match tokio::time::timeout(
subscribe_retry_delay(retry_count), Duration::from_secs(1 << retry_count.max(6)),
shutdown_rx.changed(), shutdown_rx.changed(),
) )
.await .await
@@ -62,7 +62,7 @@ pub fn spawn_broadcast_subscriber(inner: Arc<Inner>, mut shutdown_rx: watch::Rec
break; break;
} }
Err(_) => { Err(_) => {
retry_count = retry_count.saturating_add(1); retry_count += 1;
continue; continue;
} }
} }
@@ -234,11 +234,6 @@ 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 { fn log_event(event: &BroadcastEvent) -> trc::Value {
match event { match event {
BroadcastEvent::PushNotification(notification) => match notification { BroadcastEvent::PushNotification(notification) => match notification {
@@ -301,19 +296,3 @@ fn log_event(event: &BroadcastEvent) -> trc::Value {
BroadcastEvent::QueueRefresh => "QueueRefresh".into(), 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));
}
}
+20 -57
View File
@@ -567,14 +567,20 @@ 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( async fn build_tracing_span_document(
server: &Server, server: &Server,
span_id: u64, span_id: u64,
) -> trc::Result<Option<IndexDocument>> { ) -> trc::Result<Option<IndexDocument>> {
use common::telemetry::tracers::store::MaybeTrace; use common::telemetry::tracers::store::MaybeTrace;
use registry::schema::structs::Search; use registry::schema::{enums::SearchTracingField, structs::Search};
use store::write::{TelemetryClass, ValueClass}; use store::{
search::TracingSearchField,
write::{TelemetryClass, ValueClass},
};
use trc::Key;
let settings = server let settings = server
.registry() .registry()
@@ -584,6 +590,7 @@ async fn build_tracing_span_document(
if !settings.index_telemetry { if !settings.index_telemetry {
return Ok(None); return Ok(None);
} }
let wants = |field: SearchTracingField| settings.index_tracing_fields.iter().any(|f| *f == field);
let Some(MaybeTrace(Some(trace))) = server let Some(MaybeTrace(Some(trace))) = server
.tracing_store() .tracing_store()
.get_value::<MaybeTrace>(ValueKey::from(ValueClass::Telemetry(TelemetryClass::Span( .get_value::<MaybeTrace>(ValueKey::from(ValueClass::Telemetry(TelemetryClass::Span(
@@ -594,67 +601,23 @@ async fn build_tracing_span_document(
return Ok(None); return Ok(None);
}; };
Ok(Some(trace_search_document(
span_id,
&trace,
&settings
.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); 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 seen = store::ahash::AHashSet::new();
let mut queue_id_indexed = false;
for event in trace.events.iter() { for event in trace.events.iter() {
if wants(SearchTracingField::EventType) && seen.insert(event.event.as_str().to_string()) {
document.index_keyword(TracingSearchField::EventType, event.event.as_str());
}
for kv in event.key_values.iter() { for kv in event.key_values.iter() {
let text = match &kv.value { let text = match &kv.value {
TraceValue::String(v) => v.value.clone(), registry::schema::structs::TraceValue::String(v) => v.value.clone(),
TraceValue::UnsignedInt(v) => v.value.to_string(), registry::schema::structs::TraceValue::UnsignedInt(v) => v.value.to_string(),
TraceValue::IpAddr(v) => v.value.to_string(), registry::schema::structs::TraceValue::IpAddr(v) => v.value.to_string(),
_ => continue, _ => continue,
}; };
match kv.key { match kv.key {
Key::QueueId => { Key::QueueId if wants(SearchTracingField::QueueId) => {
let Ok(queue_id) = text.parse::<u64>() else { if seen.insert(format!("q:{text}")) {
continue; document.index_keyword(TracingSearchField::QueueId, &text);
};
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 Key::From
@@ -685,7 +648,7 @@ pub fn trace_search_document(
} }
} }
} }
document Ok(Some(document))
} }
// inbuxa: UD-1, UD-4: archives a deleted file, event or contact noted at // inbuxa: UD-1, UD-4: archives a deleted file, event or contact noted at
+135 -6
View File
@@ -2,6 +2,8 @@
* SPDX-FileCopyrightText: 2020 Stalwart Labs LLC <[email protected]> * SPDX-FileCopyrightText: 2020 Stalwart Labs LLC <[email protected]>
* *
* SPDX-License-Identifier: AGPL-3.0-only OR LicenseRef-SEL * SPDX-License-Identifier: AGPL-3.0-only OR LicenseRef-SEL
*
* Modified by Coffey Labs in 2026 for INBUXA.
*/ */
use crate::task_manager::acme::AcmeTask; use crate::task_manager::acme::AcmeTask;
@@ -27,6 +29,7 @@ use common::network::limiter::ConcurrencyLimiter;
use common::network::{ServerInstance, TcpAcceptor}; use common::network::{ServerInstance, TcpAcceptor};
use common::{Inner, Server}; use common::{Inner, Server};
use registry::schema::enums::TaskType; use registry::schema::enums::TaskType;
use registry::schema::prelude::ObjectType;
use registry::schema::structs::{ use registry::schema::structs::{
Task, TaskManager, TaskRetryStrategy, TaskStatus, TaskStatusFailed, TaskStatusRetry, Task, TaskManager, TaskRetryStrategy, TaskStatus, TaskStatusFailed, TaskStatusRetry,
}; };
@@ -337,6 +340,7 @@ impl TaskQueueManager for Server {
// Retrieve tasks pending to be processed // Retrieve tasks pending to be processed
let mut tasks = Vec::new(); let mut tasks = Vec::new();
let mut unreadable = Vec::new();
let now = Instant::now(); let now = Instant::now();
let mut next_event = None; let mut next_event = None;
let roles = &self.core.network.roles; let roles = &self.core.network.roles;
@@ -351,12 +355,21 @@ impl TaskQueueManager for Server {
let task_id = key.deserialize_be_u64(U64_LEN)?; let task_id = key.deserialize_be_u64(U64_LEN)?;
if task_due <= now_timestamp { if task_due <= now_timestamp {
let task_type_idx = value.deserialize_be_u16(0)?; // inbuxa: a row whose task type can't be read is
let task_type = TaskType::from_id(task_type_idx).ok_or_else(|| { // set aside, not allowed to end the scan: every
trc::StoreEvent::DataCorruption // task due after it would wait behind it
.caused_by(trc::location!()) let Some((task_type_idx, task_type)) = value
.ctx(trc::Key::Value, value) .deserialize_be_u16(0)
})?; .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 { let enabled = match task_type {
TaskType::IndexDocument TaskType::IndexDocument
| TaskType::UnindexDocument | TaskType::UnindexDocument
@@ -446,6 +459,11 @@ 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() { if !tasks.is_empty() {
trc::event!( trc::event!(
TaskManager(TaskManagerEvent::TaskAcquired), TaskManager(TaskManagerEvent::TaskAcquired),
@@ -686,3 +704,114 @@ 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,
}
}
+29 -13
View File
@@ -2,6 +2,8 @@
* SPDX-FileCopyrightText: 2020 Stalwart Labs LLC <[email protected]> * SPDX-FileCopyrightText: 2020 Stalwart Labs LLC <[email protected]>
* *
* SPDX-License-Identifier: AGPL-3.0-only OR LicenseRef-SEL * SPDX-License-Identifier: AGPL-3.0-only OR LicenseRef-SEL
*
* Modified by Coffey Labs in 2026 for INBUXA.
*/ */
use registry::{ use registry::{
@@ -40,35 +42,49 @@ pub trait InternalReportIndex: ObjectImpl {
fn primary_key(&self) -> ValueClass; 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( fn reschedule_ops(
&mut self, &mut self,
batch: &mut BatchBuilder, batch: &mut BatchBuilder,
item_id: u64, item_id: u64,
revision: u64, revision: u64,
at: UTCDateTime, at: UTCDateTime,
queued: Option<&Task>,
) { ) {
let current_deliver_at = self.deliver_at(); 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 { if current_deliver_at != at || queued_due != new_due {
let object = Self::OBJECT; let object = Self::OBJECT;
let object_id = object.to_id(); let object_id = object.to_id();
let key = ValueClass::Registry(RegistryClass::Item { object_id, item_id }); let key = ValueClass::Registry(RegistryClass::Item { object_id, item_id });
self.set_deliver_at(at); 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 batch
.assert_value(key.clone(), AssertValue::Hash(revision)) .schedule_task_with_id(item_id, self.task(item_id))
.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()); .set(key, self.to_pickled_vec());
} }
} }
+1 -1
View File
@@ -81,7 +81,7 @@ fn legacy_setting(name: &str, is_set: impl Fn(&str) -> bool) -> Option<String> {
#[macro_export] #[macro_export]
macro_rules! brand_version { macro_rules! brand_version {
() => { () => {
"2026.9.24.3" "2026.9.24.4"
}; };
} }
+3 -8
View File
@@ -212,15 +212,10 @@ unchanged.
- **MON-16.** With `indexTelemetry` on, storing a trace schedules an - **MON-16.** With `indexTelemetry` on, storing a trace schedules an
`IndexTrace` task. The task builds one document for `SearchIndex::Tracing` `IndexTrace` task. The task builds one document for `SearchIndex::Tracing`
with the fields named in `indexTracingFields`: with the fields named in `indexTracingFields`:
- `eventType`: the trace's opening event, as its numeric id; - `eventType`: every event type in the trace;
- `queueId`: the first `queueId` value, as an integer; - `queueId`: every `queueId` value;
- `keywords`: every address in `from` and `to`, each address's domain, every - `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 So searching `example.org` finds every trace to or from that domain, as the
upstream suite expects. With `indexTelemetry` off nothing is indexed, and upstream suite expects. With `indexTelemetry` off nothing is indexed, and
the `text` and `queueId` filters are refused (see "Interfaces"). the `text` and `queueId` filters are refused (see "Interfaces").
+3
View File
@@ -2,9 +2,12 @@
* SPDX-FileCopyrightText: 2020 Stalwart Labs LLC <[email protected]> * SPDX-FileCopyrightText: 2020 Stalwart Labs LLC <[email protected]>
* *
* SPDX-License-Identifier: AGPL-3.0-only OR LicenseRef-SEL * SPDX-License-Identifier: AGPL-3.0-only OR LicenseRef-SEL
*
* Modified by Coffey Labs in 2026 for INBUXA.
*/ */
pub mod analyze; pub mod analyze;
pub mod dmarc; pub mod dmarc;
pub mod reschedule; // inbuxa: report reschedules and unreadable queue rows
pub mod scheduler; pub mod scheduler;
pub mod tls; pub mod tls;
+370
View File
@@ -0,0 +1,370 @@
/*
* 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;
}
}
+5 -253
View File
@@ -2,8 +2,6 @@
* SPDX-FileCopyrightText: 2020 Stalwart Labs LLC <[email protected]> * SPDX-FileCopyrightText: 2020 Stalwart Labs LLC <[email protected]>
* *
* SPDX-License-Identifier: AGPL-3.0-only OR LicenseRef-SEL * SPDX-License-Identifier: AGPL-3.0-only OR LicenseRef-SEL
*
* Modified by Coffey Labs in 2026 for INBUXA.
*/ */
use crate::utils::{ use crate::utils::{
@@ -11,22 +9,14 @@ use crate::utils::{
server::TestServer, server::TestServer,
temp_dir::TempDir, temp_dir::TempDir,
}; };
use ::registry::schema::{ use ::registry::schema::enums::CompressionAlgo;
enums::{CompressionAlgo, TaskStoreMaintenanceType},
prelude::ObjectType,
structs::Task,
};
use ahash::AHashSet; use ahash::AHashSet;
use common::{ use common::{DATABASE_SCHEMA_VERSION, manager::backup::BackupParams};
DATABASE_SCHEMA_VERSION,
manager::{SPAM_CLASSIFIER_KEY, SPAM_TRAINER_KEY, backup::BackupParams},
};
use store::{ use store::{
rand, rand,
write::{ write::{
AnyClass, AnyKey, BatchBuilder, BlobLink, BlobOp, Operation, QueueClass, QueueEvent, AnyClass, AnyKey, BatchBuilder, BlobLink, BlobOp, Operation, QueueClass, QueueEvent,
RegistryClass, TaskQueueClass, ValueClass, RegistryClass, ValueClass, key::KeySerializer,
key::{DeserializeBigEndian, KeySerializer},
}, },
*, *,
}; };
@@ -177,50 +167,6 @@ pub async fn test(test: &TestServer) {
} }
db.write(batch.build_all()).await.unwrap(); 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 // Create directory data
println!("Creating directory data..."); println!("Creating directory data...");
let mut batch = BatchBuilder::new(); let mut batch = BatchBuilder::new();
@@ -239,17 +185,6 @@ pub async fn test(test: &TestServer) {
println!("Calculating store hash..."); println!("Calculating store hash...");
let snapshot = Snapshot::new(&db).await; let snapshot = Snapshot::new(&db).await;
assert!(!snapshot.keys.is_empty(), "Store hash counts are empty",); 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 // Export store
println!("Exporting store..."); println!("Exporting store...");
@@ -275,188 +210,22 @@ pub async fn test(test: &TestServer) {
.finalize(), .finalize(),
); );
db.write(batch.build_all()).await.unwrap(); db.write(batch.build_all()).await.unwrap();
for (key, _) in &named_blobs { test.server.core.restore(temp_dir.path.clone()).await;
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(); let mut batch = BatchBuilder::new();
batch.clear(ValueClass::NodeId(0)); batch.clear(ValueClass::NodeId(0));
db.write(batch.build_all()).await.unwrap(); 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 // Verify hash
print!("Verifying store hash..."); print!("Verifying store hash...");
snapshot.assert_is_eq(&Snapshot::new(&db).await); snapshot.assert_is_eq(&Snapshot::new(&db).await);
assert_named_blobs(test.server.blob_store(), &named_blobs).await;
println!(" GREAT SUCCESS!"); 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 // Destroy store
for (key, _) in &named_blobs {
test.server.blob_store().delete_blob(key).await.unwrap();
}
store_destroy(&db).await; store_destroy(&db).await;
store_assert_is_empty(&db, db.clone().into(), true).await; store_assert_is_empty(&db, db.clone().into(), true).await;
temp_dir.delete(); 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)] #[derive(Debug, PartialEq, Eq)]
struct Snapshot { struct Snapshot {
keys: AHashSet<KeyValue>, keys: AHashSet<KeyValue>,
@@ -471,19 +240,7 @@ struct KeyValue {
impl Snapshot { impl Snapshot {
async fn new(db: &Store) -> Self { async fn new(db: &Store) -> Self {
Self::build(db, !db.is_sql(), true).await let is_sql = db.is_sql();
}
/// 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(); let mut keys = AHashSet::new();
@@ -508,12 +265,7 @@ impl Snapshot {
(SUBSPACE_QUOTA, !is_sql), (SUBSPACE_QUOTA, !is_sql),
(SUBSPACE_REPORT_OUT, true), (SUBSPACE_REPORT_OUT, true),
(SUBSPACE_REPORT_IN, true), (SUBSPACE_REPORT_IN, true),
(SUBSPACE_DIRECTORY, true),
(SUBSPACE_INBUXA, true),
] { ] {
if subspace == SUBSPACE_BLOBS && !with_blobs {
continue;
}
let from_key = AnyKey { let from_key = AnyKey {
subspace, subspace,
key: vec![0u8], key: vec![0u8],
-163
View File
@@ -2,8 +2,6 @@
* SPDX-FileCopyrightText: 2020 Stalwart Labs LLC <[email protected]> * SPDX-FileCopyrightText: 2020 Stalwart Labs LLC <[email protected]>
* *
* SPDX-License-Identifier: AGPL-3.0-only OR LicenseRef-SEL * 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}; use crate::{store::deflate_test_resource, utils::server::TestServer};
@@ -124,10 +122,6 @@ pub async fn test(test: &TestServer) {
println!("Running global id filtering tests..."); println!("Running global id filtering tests...");
test_global(store.clone()).await; 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 // Large document insert test
println!("Running large document insert tests..."); println!("Running large document insert tests...");
let mut large_text = String::with_capacity(20 * 1024 * 1024); let mut large_text = String::with_capacity(20 * 1024 * 1024);
@@ -815,160 +809,3 @@ async fn test_global(store: SearchStore) {
AHashSet::from_iter([3, 4, 5]) 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();
}
}
-67
View File
@@ -148,73 +148,6 @@ pub async fn test(test: &mut TestServer) {
"test 9: to" "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 // Acceptance test 24: destroy removes a trace; create is refused
let trace_id = traces[0]["id"].as_str().unwrap().to_string(); let trace_id = traces[0]["id"].as_str().unwrap().to_string();
let response = admin let response = admin