Release 3.9.1

Release Date

v3.9.1 was released to the public on June 19, 2024.

New in This Release

This release introduces the following changes since version 3.8.x:

1. api-service: list volumes --name will now return the correct results when filtering by name.
2. aws: Added the option to choose the PlacementGroup policy to Lightbits cluster for deploying on AWS.
3. azure: Updated the supported regions list.
4. cluster-manager: Calculated eviction progress more accurately.
5. cluster-manager: Volume placement logic can now place one of a volume's replicas on a node in permanent failure state, rather than failing volume creation.
6. cluster-manager: Fixed a bug where a snapshot could get stuck in Deleting state due to a race between deleting a snapshot and deleting its volume.
7. cluster-manager: Fixed an issue where a volume in Unavailable state could be stuck in this state when the system tries to migrate the volume to another node.
8. cluster-manager: Fixed an issue that could have caused a volume to be stuck in Migrating state after it had been migrated from a node and back to the same node after a short time (about 30 seconds).
9. cluster-manager: When a Placement Group (PG) migration is completed successfully, the migration task is deleted. If the deletion occurs while the PG manager still has the PG cached, the PG would fail the migration, as there is no task but also no migration task. This can be fixed by taking the Rebuild state into account as well.
10. data-layer: Fixed a bug in handling deletions of snapshots that - under rare conditions - could lead to rebuilds not being started, and volumes being stuck in Degraded/Read-Only Protection states.
11. duroslight: Enabled the detection of GFTL read or write stalls. Stalls are now reported to the node-manager in the appropriate bits in the Health Check response. The node-manager follows up by turning the node inactive and auto-reviving it.
12. duroslight: Implemented recovery from SSD journal. This is part of the upcoming SSD journaling feature and is currently disabled by default.
13. lbavs: Accepted multiple values (a list) as input values for failure domains.
14. lbavs: Allowed REST APIs to authenticate and work with valid access tokens.
15. lbavs: Datastore names are kept unique across multiple Lightbits clusters to avoid conflicting names in an AVS environment.
16. lbavs: Enabled Azure AVS RunCommand to return datastore along with names of all host(s) actively accessing the datastore.
17. lbavs: Enabled the Azure AVS RunCommand to return a host along with the names of all accessible datastore(s).
18. lbavs: Enabled HTTPS (443) authentication on REST API.
19. lbavs: Enabled users to view all available storage adapter(s) for a given ESXi host.
20. lbavs: Kickoff workflow to connect host(s) to target(s) for a replaced or newly added host to an AVS cluster.
21. lbavs: Ensured that a valid value was used when resizing a datastore.
22. lbavs: Persist AVS cluster inventory information to a file to determine any changes to the cluster.
23. lbavs: Polled AVS cluster inventory to determine any changes to an AVS cluster.
24. lbavs: Supported a configurable time-based scheduler to poll the AVS cluster inventory.
25. lbcli: Added missing support for the "human-readable" output format to the get server command.
26. lbcli: lbcli list events --component-type will now validate if passed an incorrect component or no component at all.
27. lb_monitor: Added an option to show Lightbits events in local time. Added UTC time in the top header and a timezone (TZ) header for the Events section. Added line numbers for etcd, server, and node output for clarity.
28. lb-support: Added missing collections. Improved help readability. Improved output structure of the collected log file for easier and more effective analysis. Added an xz compression option with -x. Added an option to skip collecting contents of /var/log.
29. lightbits-api: Added server health indication. Server object returned by list/get server now contains a health field returning either OK, Warning, or Error.
30. lightbits-api: get and list volume operations will now return the creation time of the volume for any volumes created on Lightbits v3.9.1 or newer. Volumes that were created before v3.9.1 will return a NULL value in the new creationTime field.
31. lightbits-api: list volumes --name will now return the correct results: will show only the volumes with the name as in the filter (if no project - there might be more than one). If a volume is not available (e.g., in Creating or Deleting state) it will show in the results only if the “show all” flag is used.
32. node-manager: Fixed an extremely rare race in the service's startup flow that - combined with a very rare chain of migrations of the same volume - may in theory lead to incomplete rebuilds and eventually data loss.
33. openstack: Fixed the issue of writing double ALLOW_ANY entries to a volume's IPACL, when the IPACL is disabled via configuration file.
34. upgrade-manager: Increased the default server upgrade-able timeout to four hours, even if the default was set by a previous version to 10 or 60 minutes.
35. userlbe: Fixed a possible race between the exporter accessing some metrics and volume deletion. This race, when hit, could cause a Lightbits node to stop serving I/Os.
36. userlbe: Fixed an issue where the disk recovery process would hang if a second disk was removed after graceful recovery on an empty system.

Installation and Upgradeability

You can upgrade to this release from all previous Lightbits v3.6.x, v3.7.x, or 3.8.x releases.