Issue Type | Description | ID |
|---|
Enhancement
| data-layer: To further enhance data consistency, the synchronization process for nodes reconnecting after a network disruption or server restart/downtime has been improved. This ensures that all pending operations, such as volume and snapshot deletions, are correctly applied and all nodes are properly synchronized, maintaining the proper protection state for all volumes.
| LBM1-31156 |
Enhancement
| duroslight: Support for enabling multiple sub-systems has been added to extend the number of clients that can connect to a Lightbits cluster. This new capability directs each client to a pre-configured sub-system, allowing for a substantial increase in overall client connection capacity. Important: In this initial version, the auxiliary sub-system mode and in-band authentication are mutually exclusive and cannot be used at the same time. The ability to use both together is under active development and is planned for a future release.
| LBM1-38254 |
Enhancement
| duroslight: When the number of metrics is high, metric collection can stall the reactor, resulting in increased max/tail latencies for specific workloads. The collection of metrics will now yield so that the IO latency is not impacted.
| LBM1-38857 |
Enhancement
| Photon v1.2 includes several usability improvements, most notably: making the UI workable on small or zoomed displays with scrollbars and preserved dialogs, adding cluster UUID visibility on the Clusters page and side panel, delivering additional UX and UI refinements including fault focused color usage, ensuring that alert dates are always visible in the side panel to clearly indicate when each alert occurred, and enabling logo to Dashboard navigation - along with various fixes and internal testability improvements. | LBM1-41808 |
Major
| duroslight: Added a sanity check for rebuild/data recovery object data size. This allows older primary software sending illegal objects to persist them without triggering assertion failures.
| LBM1-38330 |
Major
| duroslight: Added a sanity check on objects read from the primary to ensure that the rebuild read response data sizes comply with the replication protocol, even if the original object is malformed. This prevents link failures caused by sendmsg: Bad Address errors on Alma9 kernels.
| LBM1-37887 |
Major
| duroslight: Contributed to data integrity and system reliability by fixing a rare issue where recent data could be reverted to the data stored in the last snapshot. This could occur if the most recent snapshot was deleted while a replica node was offline, potentially causing the new data to be reverted to the snapshot's older data during a rebuild.
| LBM1-39211 |
Major
| duroslight: Ensured that the primary adheres to protocol assumptions when passing objects, and that the secondary sanity checks received objects - detecting and correcting any violations so malformed objects can still be safely persisted. Previously, when a primary read an object requested by a secondary rebuild, it might have detected corruption (e.g., decompression failure, CRC mismatch). However, if dek_generation was corrupted (non-zero), the primary may have missed the issue and treated the object as encrypted, passing it as is. The secondary then assumed that the object was valid and writable, which could cause problems if the object was malformed.
| LBM1-37650 |
Major
| duroslight: Further improved data integrity in DCPMM environments by resolving a rare issue related to taking a snapshot after an abrupt failure. This enhancement ensures that recently written data is consistently preserved, preventing unintended reversion to prior snapshot values, even in these infrequent scenarios. | LBM1-39168 |
Major
| duroslight: To further enhance system stability, Intel® Data Streaming Accelerator (DSA) support is now disabled by default. This precautionary measure prevents potential hardware exceptions, ensuring more reliable operation.
| LBM1-39628 |
Major
| node-manager: Contributed to the overall stability of maintenance operations by resolving a rare deadlock. This fix prevents a node from hanging during shutdown if a new shutdown command is issued before a previous startup has fully completed.
| LBM1-39872 |
Major
| node-manager: To further enhance system stability, a rare case of a potential deadlock in the node's shutdown flow has been resolved. This addresses scenarios where a shutdown is initiated before the node's power-up sequence has fully completed, ensuring more robust behavior during rapid state transitions.
| LBM1-38754 |
Major
| nvme: Contributed to overall system reliability by preventing a node from sending I/O errors to a client after it had already been marked as inactive. This fixes a rare race condition, ensuring a cleaner and more stable client failover during a node failure.
| LBM1-39661 |
Major
| userlbe: GFTL allocations are now NUMA-interleaved. This prevents rare cases of single-node memory exhaustion.
| LBM1-39819 |
Minor
| ansible: The installer will no longer update system packages to the latest version, to avoid unexpected changes.
| LBM1-37738 |
Minor
| api-service: Fixed an issue in the 'list connected-hosts' command, where the IPACL was ignored when using a volume UUID filter, resulting in hosts being listed even if they were not actually connected to the volume.
| LBM1-37852 |
Minor
| api-service: Fixed an issue where - following a restart - the api-service could return a snapshot with all statistics set to zero.
| LBM1-37725 |
Minor
| api-service: Fixed an issue where - following an api-service restart - there was a brief window during which some nodes and servers might have been missed, potentially resulting in a partial list of nodes and/or servers being returned.
| LBM1-36980 |
Minor
| api-service: Resolved an API service crash that could occur when listing cluster or node statistics while they were being updated. This action enhances the stability and reliability of these operations.
| LBM1-38600 |
Minor
| api-service: The DEKsRetentionPeriod cluster configuration key has been renamed to DeksRetentionPeriod to align with camel case conversion standards, enhancing naming consistency within the API service.
| LBM1-39727 |
Minor
| cluster-manager: Extended the capability of the cluster encryption feature to support IPv6-only environments. This change resolves an IP address formatting issue, contributing to the feature's reliability and compatibility across all supported network configurations.
| LBM1-37460 |
Minor
| cluster-manager: The Cluster Manager now handles the node replacement process with greater stability and predictability. This enhancement resolves several underlying race conditions to ensure consistent and expected behavior.
| LBM1-39119 |
Minor
| cluster-manager: The TLS configuration for internal cluster communication now enforces a predefined list of allowed ciphers.
| LBM1-37337 |
Minor
| data-layer: Fixed a bug that caused the Events Manager to stop and return stale events after encountering unexpected or deleted events like EnableServerEncryptionFailed (from versions 3.12 and 3.13).
| LBM1-38043 |
Minor
| discovery-service: Fixed an issue where Asynchronous Event Notifications (AEN) from the discovery service were missing during IP Access Control List (IPACL) updates, potentially preventing a client from connecting automatically as expected.
| LBM1-39002 |
Minor
| duroslight: For more precise performance analysis, the wrlat_xmit latency metric was added. This new measurement captures the time taken to send a Ready-to-Transfer (R2T) packet, providing valuable insight into the I/O path after internal resource gating and QoS delays but before the client begins its data transmission.
| LBM1-37206 |
Minor
| duroslight: Resolved a very rare edge case that could cause the Duroslight process to hang during graceful node shutdown if the primary node is also shutdown/crashed at the same time. This enhancement specifically improves stability during concurrent rebuild operations, even when the rebuild primary node fails just moments before.
| LBM1-39356 |
Minor
| The wrlat_reply_qued statistic in Duroslight has been refined to accurately reflect only write I/O operations, excluding other NVMe commands. This improvement provides a more precise measure for analyzing write latency. | LBM1-38751 |
Minor
| System reliability has been further enhanced by resolving multiple underlying rare race conditions. This improvement ensures that the health status of each node is always reported with accuracy and that transitions between states - such as active and inactive - are handled consistently. | LBM1-38932 |
Minor
| This change ensures that firewall rules are now correctly applied for each specified IP when creating an admin-endpoint, contributing to overall security. | LBM1-39869 |
Minor
| userlbe: Contributed to data integrity and system reliability by fixing a rare issue where a volume could get stuck in a "Deleting" state. This could occur if a volume was deleted and then quickly recreated before the deletion fully completed, causing the affected volume replica to silently ignore all incoming writes. This fix ensures that the volume state is correctly handled, preventing data inconsistency.
| LBM1-40243 |
