Release 3.8.1

Release Date

v3.8.1 was released to the public on April 24, 2024.

New in This Release

This release introduces the following changes since version 3.7.x:

1. ansible: The log-collector playbook was updated to collect using the system log script. Where it's missing, it uses the uploaded one. Logs are dumped into an organized directory in /tmp on the Ansible host. For more details, see the new log collector MD file.

2. api-service: Better input checking in CreateCredential().

3. api-service: Enforce a limit of a max of 24K of volumes per single project.

4. api-service: Enforce limits of the number of snapshot policies per volume and the total number of snapshots policies. The limits are max 64K policies total and 8 policies per volume.

5. api-service: Fix the override of HTTPS “failed precondition” & "conflict" error codes, in the delete volume flow.

6. api-service: Fix a possible case of returning non gRPC errors in the delete volume flow.

7. api-service: Fix the rare case where API requests could have been handled but an error was returned to the user (a false negative).

8. api-service: Update to newer JWT libraries.

9. api-service: When enabling a server that was disabled by using the evict-data option, the LastEviction field of the server will not be reset anymore. It will retain the information on the last eviction, until the next data-eviction.

10. azure: Add an internal Lightbits vIP, link to the JWT secret inside the vault, and storage account URL

11. azure: Enable downloading troubleshooting bundles and viewing log messages in the Azure portal.

12. azure: Updated the supported regions list.

13. azure: Added support for incremental backup and restore of Lightbits volumes to/from the Azure Blob service.

14. cluster-manager: Added an automatic abort for data migrations where their target node is currently evicting data.

15. cluster-manager: Added support for aborting ongoing data migration when data eviction is requested.

16. cluster-manager: Added support for disabling data eviction. While a server evicts its data, the server state is Enabling. A user can abort the data eviction by using the command lbcli enable server. When the system is done aborting the eviction, the server state will become Enabled.

17. cluster-manager: Fixed an issue in which we could mistakenly show a snapshot as created, before its internal creation is actually finished. This can only happen when all of the snapshot's replica nodes were inactive when it was created.

18. cluster-manager: Fixed a rare bug due to which there could have been more than a single active cluster manager. This could lead to different and unexpected behaviors.

19. cluster-manager: A stop leader campaign when context was cancelled. This caused high CPU usage due to attempts to keep trying that were unsuccessful.

20. duroslight: Added stats for in-flight I/O commands.

21. duroslight: Fixed an issue that occurred when a machine has IPv6 disabled completely when binding a socket to '::1' (i.e., 127.0.0.1, localhost), and listening to it failed - causing Duroslight to fail to send its metrics to Prometheus.

22. events: Changed the severity of the ServerCreateFailed event from Info to High.

23. installation: Added a missing dependency on fio-engine-libaio on RH9 derivatives.

24. installation: Certificates for peer etcd servers previously created with the same ansible_hostname for all servers, that are now properly created with the hostname of each server.

25. installation: Fix Ansible to work for a non-root (sudoer) user.

26. lbavs: Added the ability to view all available snapshot(s) of VMFS datastores.

27. lbavs: Added support for tag and create, using REST API.

28. lbavs: Enabled configuring the deployed RunCommand version in the yaml configuration file.

29. lbavs: ESXi host(s) are automatically connected to the Lightbits cluster as part of auto scaling.

30. lbavs: ESXi host(s) are automatically connected to replaced Lightbits servers as part of auto healing.

31. lbavs: Support creating a VMFS datastore, using REST API.

32. lbavs: Support creating a snapshot of the VMFS datastore, using REST API.

33. lbavs: Support deleting a VMFS datastore, using REST API.

34. lbavs: Support disconnecting an ESXi host from targets, using REST API.

35. lbavs: Support list datastore, using REST API.

36. lbavs: Support mount and unmount VMFS datastore, using REST API.

37. lbe: Ensure that lbe will not crash in rare cases that might occur if - for some reason - the node fails to perform a graceful shutdown while there are in-flight write objects

38. lbe: Fixed a crash when flush races with shutdown.

39. lbe: Fixed a rare bug where crashing during graceful recovery could occur in case the number of MD cores increased after shutdown and data from deleted snapshots were recovered.

40. lbe: Improved rebuild time in the presence of foreground writes and/or a local rebuild.

41. lbe: We now take into account the amount of wasted bytes due to padding in extreme compression scenarios - when deciding whether to go into read-only mode

42. node-manager: Fixed a rare issue where if a node was simultaneously disconnected and encountered a data plane issue, it could get to a state that prevented itself from stopping its internal services.

43. openstack: Fixed a live migration bug in the Cinder driver, where the target server's IPACLs would overwrite the source server's IPACLs.

44. Upgrade-manager: Starting from this release, it is possible to skip one version when upgrading. v3.8.1 is the first release that can be upgraded directly from v3.6.x - rather than upgrading to 3.7.x first and only then to v3.8.1.

Installation and Upgradeability

You can upgrade to this release from all previous Lightbits v3.6.x or v3.7.x releases.