The initial RPC create is now persisted synchronously before sending
(persist-before-send); only status updates remain on the async batched
queue. The batched-insert plumbing that was retained for reversibility is
now dead code and is removed:
- remove createAsync from RpcService/RpcDao/BaseRpcService/JpaRpcDao
- delete RpcQueueEntry (insert/update discriminator no longer needed)
- rename RpcInsertRepository -> RpcUpdateRepository; the queue now carries
RpcEntity directly and batches UPDATEs only
- re-aim JpaRpcDaoTest / TbRpcServiceTest off the removed insert API
No behavior change: update batching, uuid partitioning/sort, and the
response COALESCE preservation are all unchanged.
- build saveOrUpdate result inside the transaction so updateCounts is a
non-null local taken straight from batchUpdate, avoiding an NPE on the
@Nullable transactionTemplate.execute() result
- guard HashPartitioner against a non-positive partition count
- fail fast in TbRpcService when sql.rpc.callback_threads < 1
- add a deterministic saveOrUpdate coalesced-batch test and rename the
timing-dependent async test to reflect what it actually verifies
- Rename TbRpcService.save -> update so the create/update pairing is
self-documenting (matches createAsync/updateAsync); update all actor
call sites and unit tests.
- Extract the (hash & 0x7FFFFFFF) % n striping into a shared
HashPartitioner helper used by both TbSqlBlockingQueueWrapper and
TbRpcService, so the end-to-end submission-order invariant has a single
source of truth.
- Collapse the insert/update split in RpcInsertRepository.saveOrUpdate
into a single partitioningBy pass.
- Remove the unused TbRpcService.findRpcById (and its now-orphaned import).
- Restore the prime-thread-count guidance on the RPC batch_threads /
callback_threads yml comments.
Async RPC persistence now returns a per-write Boolean: an INSERT-on-conflict always persists (true), while an UPDATE-by-id reports false when its WHERE id = ? matched no row (RPC deleted via TTL/manual delete). TbRpcService notifies the rule engine only when the write actually persisted, restoring the findById-null skip the async refactor had dropped.
Also: drop the unused tenantId param from createAsync/updateAsync; make TbRpcService's callback-thread count constructor-injectable (removes test reflection) with a comment on why callback striping exists alongside the queue partitioning; correct the RPC batch_threads yml comment; assert the Boolean contract in JpaRpcDaoTest and add a notification-suppression unit test.
Split the batched RPC persistence into a create path (INSERT ... ON CONFLICT)
and an update path (UPDATE ... WHERE id = ?). A status update for a row deleted
in the meantime (TTL cleanup / manual delete) is no longer resurrected,
restoring the old findById-null skip. Both paths COALESCE the response so a
status update that carries none no longer clobbers a previously stored one.
The create-vs-update intent is carried by a dedicated RpcQueueEntry record
(keeping RpcEntity pure data) and exposed via explicit createAsync/updateAsync
DAO methods instead of a boolean flag; insert/update batch scaffolding is
collapsed into a shared helper.
Adds DAO regression tests (re-queue keeps existing row, null-response
preservation, no resurrection of a deleted row) and TbRpcService create/update
wiring tests.
- Extract the tenant-profile fetch lock stripe count to a named constant (PR #15744)
- Trim the stale global-lock sentence from the per-tenant lock comment (PR #15744)
- Rename the rate-limit onCreate callback to onMiss and document its idempotency requirement (PR #15744)
- Reuse a single tenant profile local in update(TenantProfileUpdateResult) (PR #15744)
- Make the transport callback thread pool size configurable via transport.callback_thread_pool_size (PR #15744)
- Add a parameterized test locking the TransportLimitsType enum-to-profile-field mapping (PR #15744)
- Add device/gateway rate-limit coverage asserting update(tenantId) reaches tracked entities (PR #15744)
- Add a same-tenant fetch-dedup test and pin the cross-tenant test to distinct stripes (PR #15744)
Construct CoapServer and the LwM2M bootstrap server inside the init try block so a failure in the constructor or build() is cleaned up by the existing catch. Guard CoAP shutdown() against a null server. Add a DTLS-enabled CoAP test that covers the dtlsSessionsExecutor shutdown branch.
The defective code lives in common/transport/transport-api and is shared by all
transports (MQTT, HTTP, CoAP, LwM2M, SNMP); the production incident happened to
surface on MQTT.
On a cold tenant-profile cache (e.g. after a cache clear + restart), a device
reconnect storm could serialize the whole transport instance behind tenant-profile
resolution, saturating the callback pool and stalling the node for ~15 minutes.
Two compounding causes are addressed:
- DefaultTransportTenantProfileCache held a single process-wide ReentrantLock across
the synchronous cross-service getEntityProfile round-trip, so every tenant-profile
cache miss in the whole process was serialized one-at-a-time. Replace it with a
bounded set of per-tenant locks (Guava Striped) so different tenants resolve
concurrently while concurrent misses for the same tenant are still de-duplicated.
- DefaultTransportRateLimitService performed that blocking fetch inside
ConcurrentHashMap.computeIfAbsent's mapping function, holding a CHM bin lock across
the remote round-trip. Pre-fetch the tenant profile before computeIfAbsent so no bin
lock is held across I/O.
Also de-duplicate the four near-identical getXRateLimits methods into one generic
helper, move the per-type rate-limit getters onto the TransportLimitsType enum, and
avoid fetching the tenant profile four times in update(TenantId).
Upgrade LangChain4j to 1.15.1-TB1 and migrate GitHub Models to the
official OpenAI integration (OpenAiOfficialChatModel), enabling JSON
Schema (strict) structured output for GitHub Models and exposing JSON
Schema as a response format for Anthropic and Amazon Bedrock.
- Rename model capability flags to supportsSchemalessJsonOutput() and
supportsJsonSchemaOutput(), declared explicitly per provider
- Push response-format capability checks into TbResponseFormat and
simplify response-format handling in TbAiNode
- Remove obsolete manual JSON escaping for GitHub Models, which now
double-escaped requests under the official OpenAI SDK
- Drop the now-obsolete opennlp-tools version pin (the upgraded
LangChain4j fork ships opennlp-tools >= 2.5.9)
- Update the AI rule node UI to offer response formats per provider,
refresh available model lists, and migrate toggle options to @if
- Expand configurer tests to verify per-provider model configuration
Extend the IoT Hub install dialog so installing a rule chain can also set
it as the Default rule chain on a Device or Asset profile in one step.
The confirm step shows three actions: Cancel, Install (creates the chain
without targeting any profile), and Set for profile (opens a picker step).
The picker step has Cancel, Back, and Install — confirming with a profile
that already has a non-null defaultRuleChainId routes through a
confirm-overwrite step before replacing it. CALCULATED_FIELD keeps its
existing single-button flow unchanged.
Backend:
* RuleChainInstalledItemDescriptor gains a nullable EntityId
targetProfileId field, persisted in the existing descriptor JSON column
(no schema migration; @JsonIgnoreProperties(ignoreUnknown=true) handles
pre-existing rows).
* DefaultIotHubService.installRuleChain() now accepts SecurityUser and
JsonNode data; a new setAsDefaultRuleChain() helper applies the chain
as Default rule chain on the selected DEVICE_PROFILE or ASSET_PROFILE
via the existing tbDeviceProfileService / tbAssetProfileService save
paths (so the change shows up in the tenant audit log as a normal
profile update).
Frontend:
* iot-hub-install-dialog.component grows a per-ItemType selectEntityConfig
map, a 'confirm-overwrite' state, and the methods
installAsEntityProfileDefault, selectEntityBack, resolveOverwrite,
confirmOverwriteReplace, confirmOverwriteCancel.
* tb-entity-select gets a 512px min-width above the gt-sm breakpoint so
the picker renders at a consistent width regardless of the prompt text
length (mirrors the pattern in recipient-notification-dialog).
* New i18n keys: rule-chain-install-desc, rule-chain-install-as-default,
select-profile-for-rule-chain, rule-chain-overwrite-title,
rule-chain-overwrite-body, rule-chain-overwrite-replace.
SimpleCalculatedFieldState.formatResult cast the double result down to
int via TbUtils.toInt when "Decimals by default" was 0 (the UI default).
BigDecimal.intValue() returns only the low-order 32 bits, so a sum
above ~2.1B wrapped to a negative number (e.g. 3,980,173,734 ->
-314,793,562).
Add TbUtils.toLong(double) alongside toInt (toInt is left untouched to
preserve TBEL script behavior), switch the CF precision=0 path to it,
and add a Long branch in createResultJson so the JSON node is emitted
as a numeric long
- Group tomcat, commons-lang3 version properties under spring-boot.version
- Drop thymeleaf override (PE-only dependency, not present in CE)
- Drop lz4 plumbing: kafka-clients 3.9.2 and cassandra-all 5.0.7 now transitively ship at.yawk.lz4:lz4-java, making the Dec 2025 CVE hack obsolete
AbstractTbQueueConsumerTemplate.poll() returned emptyList() immediately
when partitions was empty, bypassing both doPoll() and the secondary
sleep guard (which is also skipped for backends that report
isLongPollingSupported()==true, e.g. Kafka). The result was a
permanent CPU-burning loop on consumers whose partition assignment
ended up empty after a rebalance cascade — observed on 26
ie-downlink-consumer threads (~244% total CPU) until container restart.
Route the empty-partition path through sleepAndReturnEmpty() so the
caller honors durationInMillis regardless of long-polling support.
Moved NioEventLoopGroup allocations into the try block so that a
constructor failure for the second group no longer leaks the first.
Channel close failures during cleanup now attach via addSuppressed
instead of replacing the original BindException. Narrowed the outer
catch from Throwable to Exception, removing the brittle (Error) cast
that would have masked any direct Throwable subclass.
* Feature/iot hub alarm rules (#15539)
* feat(iot-hub): scaffold ALARM_RULE item type and CREATOR_VISIBLE_ITEM_TYPES
* feat(iot-hub): wire ALARM_RULE through browse, item card, detail dialog, installed items, and install dialog
* feat(iot-hub): reorder home cards, hide Dashboards, add Alarm Rules
* feat(iot-hub): scaffold ALARM_RULE descriptor and reject install with v4.3 upgrade hint
Registers AlarmRuleInstalledItemDescriptor in the Jackson polymorphism so
4.2 can browse Alarm Rule items from the marketplace without descriptor
deserialization failures.
The install handler explicitly rejects ALARM_RULE with a friendly message
asking the user to upgrade to v4.3+. The full install/update/delete
implementation depends on CalculatedFieldType.ALARM (added in v4.3) and
will land once master is merged into this branch.
* feat(iot-hub): hide redundant Type filter on Alarm Rules browse
* feat(iot-hub): show v4.3 update info dialog directly when installing an Alarm Rule
* feat(iot-hub): implement ALARM_RULE install / update / delete handlers
Replaces the temporary v4.3-upgrade-required rejection with the actual
backend handlers. Alarm rules install as CalculatedField entities with
type=ALARM, mirroring the CALCULATED_FIELD path:
- installAlarmRule parses the alarm-rule JSON, validates type==ALARM,
saves via tbCalculatedFieldService, and returns
AlarmRuleInstalledItemDescriptor with calculatedFieldId + entityId.
- updateAlarmRule re-saves the existing CF with new name, type, and
configuration.
- Delete branch removes the underlying calculated field on installed-item
delete.
- Checksum dispatch unifies CF and ALARM_RULE through a shared
calculateCalculatedFieldChecksum(CalculatedFieldId) helper.
Frontend: drops the v4.3 info-dialog interception in IotHubActionsService
(install now actually runs); restores the select-entity branch for
ALARM_RULE in the install dialog so users can pick a parent device.
Removes the now-unused alarm-rule-install-update-required* locale keys.
* feat(iot-hub): scaffold ALARM_RULE item type and CREATOR_VISIBLE_ITEM_TYPES
* feat(iot-hub): wire ALARM_RULE through browse, item card, detail dialog, installed items, and install dialog
* feat(iot-hub): reorder home cards, hide Dashboards, add Alarm Rules
* feat(iot-hub): scaffold ALARM_RULE descriptor and reject install with v4.3 upgrade hint
Registers AlarmRuleInstalledItemDescriptor in the Jackson polymorphism so
4.2 can browse Alarm Rule items from the marketplace without descriptor
deserialization failures.
The install handler explicitly rejects ALARM_RULE with a friendly message
asking the user to upgrade to v4.3+. The full install/update/delete
implementation depends on CalculatedFieldType.ALARM (added in v4.3) and
will land once master is merged into this branch.
* feat(iot-hub): hide redundant Type filter on Alarm Rules browse
* feat(iot-hub): show v4.3 update info dialog directly when installing an Alarm Rule
- Introduce HasAppliedToEntity interface used by CreatedAlarmRuleInfo and
CreatedCalculatedFieldInfo to share the entity page-link logic and
cover DEVICE / ASSET in addition to the profile types.
- Add static from(EntityId, name, CalculatedField) factories to both
records; SolutionInstallContext now uses them and detects ALARM
calculated fields via type instead of the previous TODO/hardcoded
false.
- AlarmSeverity and CalculatedFieldType expose display names used when
formatting created-alarm-rule severities and CF type column.
- DefaultSolutionService switches the CF arguments check from
BaseCalculatedFieldConfiguration to ArgumentsBasedCalculatedFieldConfiguration
and throws ThingsboardRuntimeException for missing references.
SolutionTemplateInstalledItemDescriptor and SolutionInstallResponse now
carry tenantTelemetryKeys and tenantAttributeKeys lists. SolutionService
.deleteSolution takes the full descriptor (instead of just the
created-entity list) so uninstall can also clean up tenant-scoped
telemetry/attributes. Update IotHub install descriptor population and
delete call sites; force-update path logs and continues if delete throws.
PathChildrenCache invoked EnsureContainers internally during start(), so
the parent znode existed when publishCurrentServer ran. CuratorCache does
not, so on a fresh ZK creatingParentsIfNeeded reaches ZKPaths.mkdirs with
the trailing-slash forPath argument and fails with
"Path must not end with / character", aborting Spring startup.
dshvaika
Viacheslav Klimov
Nikita Mazurenko
Andrii Landiak
Oleksandra Matviienko
dashevchenko
Dmytro Skarzhynets
Volodymyr Babak
Andrew Shvayka
Igor Kulikov