kadalu

The Kadalu team is happy to announce a new version of Kadalu Storage, 0.8.5

• Added support for PV expansion provisioned from external Volumes.
• New Prometheus metrics added. kadalu_storage_pv_capacity_bytes, kadalu_storage_pv_capacity_used_bytes and kadalu_storage_pv_capacity_free_bytes
• Prometheus annotations added to automatically configure scraping.
• Fixed issue with Python dependencies in Arm builds.
• To avoid applications loosing access to the storage after upgrade, CSI node plugins are not upgraded as part of Operator and server pods upgrade.
• Fixed issues while setting external Quota when multiple hosts provided.

• Fixed accounting issues related to Quota feature.
• Fixed issues while mounting the devices after creating fs
• Removed default storage classes created for each Volume types.
• Added one storage class for each Kadalu Storage.
• Operator: Added retry for Protocol error while watching the CRD.
• Fixed issue while deleting a Kadalu Storage when no Pvs present.

• Improvements to the External GlusterFS Volumes support.
• Added support for Disperse Storage Type.
• Added support for archiving the PVCs on delete.
• kubectl-kadalu CLI enhanced to handle kadalu format.
• kubectl-kadalu CLI enhanced to handle PVCs archive.
• Fixed Armv7 kubectl binary issue.
• Added support for Volume ID in Storage config to
  enable reusing/migrating the Storage backend.
• Fixed issue with listing Storages and deleting external volumes
• Added support for decommisioning Storage units.
• Added support for Distributed Volumes for external.
  Volumes by enhancing quota integration over ssh.
• Enabled migration support from Replica1 to Replica2/3.

• Fix some issues on arm64 arch! (fixes in glusterfs repo).
• allow support for 'decommission' of a brick (remove-brick)
• replica1 -> replica2/3 support (or the other way).
• support glusterfs-quota support in External mode.
• add support to import a volume into kadalu with volume-id option.

• Mainly building for #531 (on aarch64) bug validation. (fix merged in series_1 branch).
• check if one of the issues seen in script is fixed or not (#565)

• Standardize usage of 'External' volumes
• Disperse volume support
• Archive support for PV (through pvReclaimPolicy)
• fix the kubectl platform type issue with arm/v7

• Enhanced provisioner to check free space before provisioning from External Storage.
• I/O tests added as part of CI.
• CSI Sanity tests added as part of CI.
• CSI "ValidateVolumeCapability" API is implemented.
• CSI "ListVolumes" API is implemented.
• Fixed an issue while creating a PV with same name but different size.
• Documentation format is changed to Asciidoc.

This is a test release to check the linux/arm64 and linux/arm/v7 builds with 0.8.x series.

Also expected to reduce the image sizes further.

NOTE: not recommended to upgrade to the release.

This is an upgraded release in the 0.8 series of kadalu project.

Few things to highlight:

Platform

• linux/arm64 and linux/arm/v7 are still not enabled.

• If you need to run same, please use 0.7.7 version.
• Users who want to experiment latest code requested to run builds locally on `arm/v7' arch.
• For users who need linux/arm64 and want to experiment 0.8.x series can try arm tag of kadalu images.

glusterfs

• glusterfs version moved to series_1 branch of kadalu/glusterfs. Includes many stability fixes, including few in replicate translator.

kadalu

• no need to run quotad in server brick pods anymore. Reduces confusing logs
• fix the kubectl kadalu status command, which failed with missing sqlite3 package on server
• add the error which poped up due to missing permission to execute the kubectl command from inside the pod.
• reduce some excessive glusterfs logs by providing proper options.
• allow self-heal of files from client mounts also in Replica2/Replica3 cases, so any issues in shd daemon doesn't become blocker.

General Note: If you are upgrading from 0.7.x series, team recommends to wait till one more version. Mainly due to multiple changes which were done in 0.8 series. All new users (or new setups) and users of 0.8.0 version are advised to upgrade.

Few things which changes with this release: (compared to 0.7.x series)

Platform Note:

• linux/arm64 and linux/arm/v7 platform is not supported in this release.

Other details:

• use kadalu-storage (glusterfs fork of kadalu, which adds features like simple-quota etc).
• changes in templates to accomodate distribute volumes
• remove usage of -oprjquota on mount
• use simple quota limit feature to set the quota
• simplify the brick volfile generation, and make it uniform across
• provide 'client-pid' to clients mounted in csi controller (for simple-quota)
• fix the template files
• include quota-crawler with provisioner
• Added support for getting logs of all pods in kadalu namespace with one command (for debug) - kubectl kadalu logs
• Added support for self heal info into CLI - can be checked with kubectl kadalu healinfo
• Added support for prometheus monitoring, so all of kadalu's pods can send crucial metrics to prometheus server.
• Fixed the bug in helm chart clusterole, which caused errors while deploying the CSI resizer pods.
• Added helm chart package into release binaries. (credits @leelavg)

So what are the benefits for users:

• Instead of creating multiple host volumes of same type, they can use just one volume with all bricks
  • Scalling of the host volume is just adding storage to config and calling kubectl apply -f again.
• Any backend filesystem which supports extended attributes can be used on devices.
• no requirement of 'quotad' process

** General Note **: Team recommends users who have successfully running kadalu instance with 0.7.x version to not yet upgrade to this release, mainly as there are multiple changes in this release.

Few things which changes with this release: (compared to 0.7.x series)

• use kadalu-storage (glusterfs fork of kadalu, which adds features like simple-quota etc).
• changes in templates to accomodate distribute volumes
• remove usage of -oprjquota on mount
• use simple quota limit feature to set the quota
• simplify the brick volfile generation, and make it uniform across
• provide 'client-pid' to clients mounted in csi controller (for simple-quota)
• fix the template files
• include quota-crawler with provisioner
• Added support for getting logs of all pods in kadalu namespace with one command (for debug) - kubectl kadalu logs
• Added support for self heal info into CLI - can be checked with kubectl kadalu healinfo
• Added support for prometheus monitoring, so all of kadalu's pods can send crucial metrics to prometheus server.
• Fixed the bug in helm chart clusterole, which caused errors while deploying the CSI resizer pods.
• Added helm chart package into release binaries. (credits @leelavg)

So what are the benefits for users:

• Instead of creating multiple host volumes of same type, they can use just one volume with all bricks
  • Scalling of the host volume is just adding storage to config and calling kubectl apply -f again.
• Any backend filesystem which supports extended attributes can be used on devices.
• no requirement of 'quotad' process

Few things which changes with this release: (compared to 0.7.x series)

• use kadalu-storage (glusterfs fork of kadalu, which adds features like simple-quota etc).
• changes in templates to accomodate distribute volumes
• remove usage of -oprjquota on mount
• use simple quota limit feature to set the quota
• simplify the brick volfile generation, and make it uniform across
• provide 'client-pid' to clients mounted in csi controller (for simple-quota)
• fix the template files
• include quota-crawler with provisioner
• Added support for getting logs of all pods in kadalu namespace with one command (for debug) - kubectl kadalu logs
• Added support for self heal info into CLI - can be checked with kubectl kadalu healinfo
• Added support for prometheus monitoring, so all of kadalu's pods can send crucial metrics to prometheus server.
• Fixed the bug in helm chart clusterole, which caused errors while deploying the CSI resizer pods.
• Added helm chart package into release binaries. (credits @leelavg)

So what are the benefits for users:

• Instead of creating multiple host volumes of same type, they can use just one volume with all bricks
  • Scalling of the host volume is just adding storage to config and calling kubectl apply -f again.
• Any backend filesystem which supports extended attributes can be used on devices.
• no requirement of 'quotad' process

Few things which changes with this release:

• use kadalu-storage (glusterfs fork of kadalu, which adds features like simple-quota etc).
• changes in templates to accomodate distribute volumes
• remove usage of -oprjquota on mount
• use simple quota limit feature to set the quota
• simplify the brick volfile generation, and make it uniform across
• provide 'client-pid' to clients mounted in csi controller (for simple-quota)
• fix the template files
• include quota-crawler with provisioner
• Added support for getting logs of all pods in kadalu namespace with one command (for debug) - kubectl kadalu logs
• Added support for self heal info into CLI - can be checked with kubectl kadalu healinfo
• Added support for prometheus monitoring, so all of kadalu's pods can send crucial metrics to prometheus server.
• Fixed the bug in helm chart clusterole, which caused errors while deploying the CSI resizer pods.
• Added helm chart package into release binaries. (credits @leelavg)

So what are the benefits for users:

• Instead of creating multiple host volumes of same type, they can use just one volume with all bricks
  • Scalling of the host volume is just adding storage to config and calling kubectl apply -f again.
• Any backend filesystem which supports extended attributes can be used on devices.
• no requirement of 'quotad' process

Few things which changes with this release:

• use kadalu-storage (glusterfs fork of kadalu, which adds features like simple-quota etc).
• changes in templates to accomodate distribute volumes
• remove usage of -oprjquota on mount
• use simple quota limit feature to set the quota
• simplify the brick volfile generation, and make it uniform across
• provide 'client-pid' to clients mounted in csi controller (for simple-quota)
• fix the template files
• include quota-crawler with provisioner
• Added support for getting logs of all pods in kadalu namespace with one command (for debug) - kubectl kadalu logs
• Added support for self heal info into CLI - can be checked with kubectl kadalu healinfo
• Added support for prometheus monitoring, so all of kadalu's pods can send crucial metrics to prometheus server.
• Fixed the bug in helm chart clusterole, which caused errors while deploying the CSI resizer pods.

So what are the benefits for users:

• Instead of creating multiple host volumes of same type, they can use just one volume with all bricks
  • Scalling of the host volume is just adding storage to config and calling kubectl apply -f again.
• Any backend filesystem which supports extended attributes can be used on devices.
• no requirement of 'quotad' process

Few things which changes with this release:

• use kadalu-storage (glusterfs fork of kadalu, which adds features like simple-quota etc).
• changes in templates to accomodate distribute volumes
• remove usage of -oprjquota on mount
• use simple quota limit feature to set the quota
• simplify the brick volfile generation, and make it uniform across
• provide 'client-pid' to clients mounted in csi controller (for simple-quota)
• fix the template files
• include quota-crawler with provisioner
• Added support for self heal info into CLI - can be checked with kubectl kadalu healinfo
• Added support for prometheus monitoring, so all of kadalu's pods can send crucial metrics to prometheus server.

So what are the benefits for users:

• Instead of creating multiple host volumes of same type, they can use just one volume with all bricks
  • Scalling of the host volume is just adding storage to config and calling kubectl apply -f again.
• Any backend filesystem which supports extended attributes can be used on devices.
• no requirement of 'quotad' process

Few things which changes with this release:

* use kadalu-storage (glusterfs fork of kadalu, which adds features like simple-quota etc).
* changes in templates to accomodate distribute volumes
* remove usage of -oprjquota on mount
* use simple quota limit feature to set the quota
* simplify the brick volfile generation, and make it uniform across
* provide 'client-pid' to clients mounted in csi controller (for simple-quota)
* fix the template files
* include quota-crawler with provisioner
* Added support for self heal info into CLI - can be checked with `kubectl kadalu healinfo`
* Added support for prometheus monitoring, so all of kadalu's pods can send crucial metrics to prometheus server.

So what are the benefits for users:

* Instead of creating multiple host volumes of same type, they can use just one volume with all bricks
  - Scalling of the host volume is just adding storage to config and calling `kubectl apply -f` again.
* Any backend filesystem which supports extended attributes can be used on devices.
* no requirement of 'quotad' process

This is a minor release of kadalu, with below improvements over 0.7.6. With this release, glusterfs used in kadalu pods are now glusterfs-8 series, instead of previous glusterfs-7 (7.2 to be precise) earlier. Which fixes multiple issues with glusterfs, and is currently maintained series in glusterfs.

Specific Bug fixes:

c4b49e0 cli: Fix parsing the device or path list (#428) - Fixed #427
c453713 server: try performing mkfs only if mount fails (#420) - Fixed #419
c7ab07e cli: use the release version as default while using release package (#421) - Fixed #418
d70fffe csi: handle the case where 'kadalu' storageclass is used (#416) - Fixed #400, #399

Generic Improvements:

61d16c4 Install latest version of GlusterFS (#430)
06206d5 tests: check for 2 success in external PVC type (#422)

With this release, kadalu project brings many improvements, and few beta features.

• PV resize support (#405)
• Upgrade support (#411)
• Support for arm/v7 platform
• Support 'backup-volfile-server' feature for external gluster cluster.

And also it adds few minor fixes:

• CRD type warning fix, and also provide schema in operator file (#413)
• Fix few logging issues (#414)

Do give it a try and let us know the feedback.

This is a minor release in 0.7.x series with few major enhancements since 0.7.4

• Fixes some of the errors seen with 0.7.4 deployments
• Removes the arm64 bits from the tag
• Reduces the size of images significantly (eg. 0.7.2 operator image was 592MB, where as this version would be ~85MB)

This is a minor release with few important long term changes:

• kadalu container's base changed from fedora to ubuntu (mainly to LTS) (#385)
• a critical breaking change in kubelet issue introduced in 1.20.x of k8s is fixed (#381)
• a primitive version of helm chart support added (#375)

also 2 minor issues fixed from that of previous minor version: #376 & #359