Skip to content

Latest commit

 

History

History
179 lines (149 loc) · 6.78 KB

CHANGELOG.md

File metadata and controls

179 lines (149 loc) · 6.78 KB

Backup as a service

All notable changes to this project will be documented in this file.

The format is based on Keep a Changelog and this project adheres to Semantic Versioning.

[v0.1.10] 2020-05-27

Changed

  • Split up documentation into smaller chapters

Fixed

  • Race condition in subscriber, causing panics

[v0.1.9] 2020-05-04

Added

  • The operator passes the configured annotation names to Wrestic

Fixed

  • KeepJobs value of sub items in Schedule are not overridden

[v0.1.8] 2020-03-03

The annotations for including/excluding PVCs as well as the backupcommand annotation have changed. If you're using them this is a breaking change and needs to be adjusted properly.

Changed

  • Change defaults for BACKUP_ANNOTATION and BACKUP_BACKUPCOMMANDANNOTATION to use k8up.syn.tools as their prefix.

Fixed

  • Race conditions in the locking logic

[v0.1.7] 2020-01-03

Fixed

  • Archive cleanup panic, the archives were cleaned up twice which lead to a nilpointer

[v0.1.6] 2019-11-14

Changed

  • Simplyfied merging logic for backend configs using mergo

Added

  • More monitoring metrics

Fixed

  • PVC restores now work correctly

[v0.1.5]

Fixed

  • Memory leak due to incorrectly unregistering jobs

Added

  • Ability to define podtemplates that will be started before the actual backup is run

Fixed

  • Bug that prevents from overriding the global statsURL -> need to be fixed properly though (APPU-1673)

v0.1.4 - 2019-03-06

From now on the K8up operator uses go modules instead of dep for dependency management. Also this version needs at least v0.0.10 of wrestic.

Added

  • Ability to set file extension for stdin jobs via annotation
  • Aggregated cluster roles for users
  • More granular S3 merge of credentials and buckets (global vs. CRD definitions)

Fixed

  • Wrong restart policy for restore
  • Fix detection of failed jobs
  • Locking considers chronology of the jobs
  • Restore job handling

Changed

  • TraviCI instead of GitlabCI

Removed

  • Pod lockups from the operator

v0.1.3 - 2018-12-09

Changed

  • Don't quote backup commands
  • Observe jobs instead of pods

Fixed

  • Fix owner reference creation
  • Choose the right target for archival jobs

v0.1.2 - 2018-11-08

Changed

  • Removed a lot of duplicated code

Added

  • Locking management, jobs that disturb each other shouldn't run concurrently anymore

v0.1.1 - 2018-11-02

Fixed

  • Cleanup of the CRDs

v0.1.0 - 2018-11-01

With this release it is possible to trigger every action on demand. But there's also a schedule CRD that can put everything on its own schedule. These changes are the plumbing for further improvements especially with the locking problem.

CAUTION: All the CRD definitions have changed slightly. So this is not backwards compatible with the old CRDs.

Also some known issues in this version:

  • No lock management yet, only the plumbing, so if there's a shared repository it can and will sometimes fail due to stale locks
  • If a job still fails after it reached the back-off limit the operator doesn't notice it. That's due to the k8s job controller cleaning up the pod. This can be resolved by observing the job states.

Added

  • Archive function
  • Check function
  • Prune function
  • Schedule CRD to schedule all functions

Changed

  • Group name of the API, it's now backup.appuio.ch
  • New CRDs please consult manifest/examples for how to use them

v0.0.10 - 2018-09-28

Added

  • Pass arbitrary URL to wrestic, which is used to post additional backup information.
  • When setting the backup annotation on PVC it will be backed up even if RWO

Fixed

  • Broken locking when using global backend

v0.0.9 - 2018-09-18

Added

  • Global S3 endpoint, S3 bucket and keepJobs. Using env vars, global default values can be specified. The defaults can be overwritten by a backup spec.

v0.0.8 - 2018-09-14

Added

  • Push images to Docker Hub
  • Divide installation manifests into own files

v0.0.7 - 2018-09-13

Added

  • Global repo password. The operator can use a global password for Restic repositories for backups which don't specify an own password.

v0.0.6 - 2018-09-12

Added

  • Manage service accounts, roles and rolebindings in namespaces for backup commands. Please see the README chapter Installation changes about the new permissions necessary for the operator service account.
  • Global cluster state. The operator now knows when a prune job is running to a shared repository and waits before starting the next backup/prune.
  • Global S3 credentials. The operator can use global S3 credentials to be used with backups which don't specify own credentials.

Fixed

  • Ignore if the service account, roles or rilebindings already exist in the namespace.

v0.0.5 - 2018-08-15

Added

  • Consistent backups -> it is now possible to set backup commands on pods. This changes the prerequisites, please consult the README chapter Installation changes for more infos.

Fixed

  • Crash when a running job got deleted
  • Skip backup if there are no backup commands and no PVCs

v0.0.4 - 2018-08-06

Changes

  • Skip backup if no suitable PVCs are found
  • Minor refactoring
  • When the backup definition is deleted it will also cleanup all jobs and pods

v0.0.3 - 2018-07-27

Changed

  • CRD status, start and end are now timestamps
  • Print namespace wenn a backup is created

v0.0.2 - 2018-07-27

Added

  • Vscode launch config
  • Read credentials and password from OpenShift secrets
  • Specifiy default Prometheus push gateway for the operator
  • Read the Prometheus push gateway url from the CRD, has precedence over the operator settings

Changed

v0.0.1

Changed

  • Initial version