Releases: bitpoke/mysql-operator
0.3.1
0.3.0
Changes since version v0.2.10: list of commits
New in v0.3.0
:
- add a new cluster condition
FailoverInProgress
that marks the cluster during a failover - merge Orchestrator chart with the mysql-operator chart. Now instead of deployment it uses a statefulset.
- fix readiness probe for MySQL container (017acf9)
- set orchestrator related events on the cluster:
OrcFailureDetection
,OrcPostUnsuccessfulFailover
,OrcPostMasterFailover
,OrcPostIntermediateMasterFailover
. (80e1ccd) - nodes are removed from cluster status at scale down (027537c)
- fix remote storage delete bug (470365f)
- fix mysqlbackup deletion failure when cluster is not found (04c440a)
- fix #350 where no backups were created when one fails (24e7b81)
- fix #107, the orchestrator RAFT issue (a81233c)
- use init container for MySQL initialization (#342)
- enhance Backup Job Pod for Workload Identity (#366)
- new command line flag
--mysql-versions-to-image
to allow user to specify image for a mysql version (ffb0a22) - add print column in mysqlcluster CRD:
Ready
,Replicas
,Age
Also in v0.3.0-rc.1
:
- refactor of how the information flow works: from k8s -> Operator -> Orchestrator; the sidecar container does not connect to Orchestrator anymore.
- cleanup of deprecated fields:
bucketURI
and top-levelMySQLCluster.spec.volumeSpec
PVC specification (d909ab9, df80b28) - rename of
initBackupURI
toinitBackupURL
(a3c6556) - use of Percona CentOS based images (#254)
- don't run as
root
user in containers (#291) - rename orchestrator finalizer (to block cluster deletion while it's registered into Orchestrator) from
OrchestratorFinalizer
tomysql.presslabs.org/registered-in-orchestrator
(bfe4646) - improvement of
getBackupCandidate
function (9ce4e68) - allow specifying more PodSpec on MySQL cluster (9b6b46f)
- configure MySQL using
init-file
(beb41ce, e5823cb) - add a node controller for MySQL configuration (1950812)
- add "standard" labels on services created by the operator (#299)
- rename
AWS_SECRET_KEY
field from bucket/backup secret toAWS_SECRET_ACCESS_KEY
(#301) - use an internal status table to store the MySQL status (181909f)
- use a common headless service for all MySQL nodes to reduce the host-name length (#246)
NOTE: see upgrade page for more information about upgrading from an older version.
v0.3.0-rc.1
Changes since version v0.2.8: list of commits
The release highlights are:
-
refactor of how the information flow works: from k8s -> Operator -> Orchestrator; the sidecar container does not connect to Orchestrator anymore.
-
cleanup of deprecated fields:
bucketURI
and top-levelMySQLCluster.spec.volumeSpec
PVC specification (d909ab9, df80b28) -
rename of
initBackupURI
toinitBackupURL
(a3c6556) -
use of Percona CentOS based images (#254)
-
don't run as
root
user in containers (#291) -
rename orchestrator finalizer (to block cluster deletion while it's registered into Orchestrator) from
OrchestratorFinalizer
tomysql.presslabs.org/registered-in-orchestrator
(bfe4646) -
improvement of
getBackupCandidate
function (9ce4e68) -
allow specifying more PodSpec on MySQL cluster (9b6b46f)
-
add a node controller for MySQL configuration (1950812)
-
add "standard" labels on services created by the operator (#299)
-
rename
AWS_SECRET_KEY
field from bucket/backup secret toAWS_SECRET_ACCESS_KEY
(#301) -
use an internal status table to store the MySQL status (181909f)
-
use a common headless service for all MySQL nodes to reduce the host-name length (#246)
NOTE: see upgrade page for more information about upgrading from an older version.
0.2.7
Changelog:
- Fix cloning when using
emptyDir
and node-0 is not master but is restarted. (#250, #252 ) (thanks @pedep ) - Backup URL on remote storage, now contains the backup name, not the cluster name (#253, #257 ) (thanks @ballad89 )
- Add
remoteDeletionPolicy
on backup resource, now backups can be removed from remote storage (#249, #81 ) - Correct azure storage rclone config (#261) (thanks @jgleonard )
- Improve the backup scheduler lock (#255, #85 )
- Fix corrupted backups that where taken under cluster stress (#260, #256 )
- Fix metrics exporter user host restriction (#265 )
0.2.6
0.2.5
The main changes and features of this release are:
- Use Percona docker images based on CentOS, this change will allow using any custom docker image based on Percona server (#200, #224 )
- Fix broken command line flag:
--image-pull-secret
(#232) - Remove unused flags:
--job-grace-time
,--http-serve-addr
(#232) - Add
--namespace
flag to limit operator activity to a certain namespace (#232) - Configure MySQL to set binlog size limits by setting:
max-binlog-size
andbinlog-space-limit
(#235 ) - Add a Helm chart to deploy a MySQL cluster easily (#236)
- Development: Make Skaffold usage more friendly (#228 )
0.2.4
This release fixes some bugs related to how Orchestrator does topology recovery. The list of changes is presented below:
- Keep last known state when orchestrator is down (#219), this will prevent master service to go down when Orchestrator is not available (#203);
- Cascade deletion of the cluster now will not delete all the resources (#221, #225);
- Update controller-runtime which implies using
klog
instead ofglog
(#227); - Let the operator set the master writable (#222, #199) so now the operator does not set nodes in maintenance to prevent failover when the cluster is in read-only. Also, unwanted failovers, which can bring the cluster in co-master topology (#207) are prevented;
- Change what means for a cluster to be
Ready
, now the cluster is ready if all pods in StatefulSet areRunning
and replicating and the cluster has a master. (f85e674); - Exclude init table from backup to allow bootstrap detection on slave (0dbcc88).
v0.2.3
In this release we fixed some bugs, add some new features and add documentation.
- Fixed metrics exporter password generation to not contain
?
(#195 ) - Increase
initialDelaySeconds
to60
seconds (#184 ) - Fix recurrent backups for cluster and
backupScheduleJobsHistoryLimit
field (#182 ) - Use Kubernetes recommended labels (#132 )
- Fix query for lag detection (#174 )
- Upgrade controller-runtime deps also client-go was updated (#191 )
- Better slave initialization that from xtrabackup (#202 )
- Add support for
emptyDir
andhostPath
and changevolumeSpec
by deprecating inline specification for PVC and addpersistentVolumeClaim
as subfield (#114 ) - Add documentation (#92, #168, #170 )
v0.2.2
This release has some bug fixes and improvements for backups and for PVCs.
- Cluster will not be set as owner reference for the credentials secret. (2bed433)
- More e2e test for PVCs cleanup, the PVC for pod 0 will not be deleted. (#178 )
- Map mysqlVersion to kown image tags (dcacd6a, #171 )
- Add a new field
.spec.image
on the MysqlCluster resource for specifying the image (#171 ) - Ensure sidecar pods limits. (466ce85)
- Don't install CRDs when the CRD is already registered. (2a49e6d)
- Add e2e test for backups and fix recurrent backups (fc7735b, adb093d)
v0.2.1
This release has a new rewrite from version v0.1.x
, the operator is based on kubebuilder as underlying SDK for writing k8s operators. This version contains all previous v0.1.x
features, stability improvments, and minor bug fixing.
- Fixes bug when cluster is scaled down (#160 )
- Provides a way to delete PVC when cluster is deleted ( #73 ) ( thanks to @smanpathak )
- The PVCs has as owner reference the mysql cluster ( #79 )