PD failover #74
Merged
PD failover #74
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
4 tasks
tennix
reviewed
Sep 3, 2018
pkg/manager/member/failover.go
Outdated
| // Failover implements the logic for pd/tikv/tidb's failover and recovery. | ||
| type Failover interface { | ||
| Failover(*v1alpha1.TidbCluster) error | ||
| Recovery(*v1alpha1.TidbCluster) |
pkg/manager/member/failover.go
Outdated
| // mark a peer as failure | ||
| for podName, pdMember := range tc.Status.PD.Members { | ||
| deadline := pdMember.LastTransitionTime.Add(pf.pdFailoverPeriod) | ||
| _, exist := tc.Status.PD.FailureMembers[podName] |
pkg/manager/member/failover.go
Outdated
| deadline := pdMember.LastTransitionTime.Add(pf.pdFailoverPeriod) | ||
| _, exist := tc.Status.PD.FailureMembers[podName] | ||
| if !pdMember.Health && time.Now().After(deadline) && !exist && notDeletedCount == 0 { | ||
| if !inQuorum { |
There was a problem hiding this comment.
This can be moved to the top level immediately after inQuorum is computed.
pkg/manager/member/failover.go
Outdated
| for podName, pdMember := range tc.Status.PD.Members { | ||
| deadline := pdMember.LastTransitionTime.Add(pf.pdFailoverPeriod) | ||
| _, exist := tc.Status.PD.FailureMembers[podName] | ||
| if !pdMember.Health && time.Now().After(deadline) && !exist && notDeletedCount == 0 { |
There was a problem hiding this comment.
notDeletedCount can be moved outside of the for loop, it only needs to be checked once.
pkg/manager/member/failover.go
Outdated
| } | ||
| inQuorum := healthCount > int(tc.Spec.PD.Replicas/2) | ||
|
|
||
| notDeletedCount := 0 |
There was a problem hiding this comment.
From the code, I guess this is used to indicate that we can only failover one at a time. So it's better to add a comment for this.
| if tc.Status.PD.FailureMembers != nil { | ||
| pmm.pdFailover.Recovery(tc) | ||
| } | ||
| } else if int(tc.Spec.PD.Replicas) == int(tc.Status.PD.StatefulSet.Replicas) { |
There was a problem hiding this comment.
Unnecessary to cast replicas to int, they're both int32.
pkg/manager/meta/meta_manager.go
Outdated
| @@ -74,8 +74,8 @@ func (pmm *metaManager) Sync(tc *v1alpha1.TidbCluster) error { | |||
| } | |||
| // update meta info for pvc | |||
| pvc, err := pmm.resolvePVCFromPod(pod) | |||
There was a problem hiding this comment.
err defined but not used. Replace it with _ if you don't want to check it.
| @@ -104,6 +104,8 @@ func (pmm *metaManager) resolvePVCFromPod(pod *corev1.Pod) (*corev1.PersistentVo | |||
| pvcName = vol.PersistentVolumeClaim.ClaimName | |||
| break | |||
| } | |||
| case v1alpha1.TiDBMemberType.String(): | |||
There was a problem hiding this comment.
Should make a default for TiDB and others.
| @@ -35,6 +35,12 @@ spec: | |||
| - /usr/local/bin/tidb-controller-manager | |||
| - -default-storage-class-name={{ .Values.defaultStorageClassName }} | |||
| - -cluster-scoped={{ .Values.clusterScoped }} | |||
| {{- if .Values.controllerManager.autoFailover }} | |||
| - -auto-failover={{ .Values.controllerManager.autoFailover }} | |||
There was a problem hiding this comment.
Use {{ .Values.controllerManager.autoFailover | default false }}.
| - -auto-failover={{ .Values.controllerManager.autoFailover }} | ||
| {{- end }} | ||
| {{- if .Values.controllerManager.pdFailoverPeriod }} | ||
| - -pd-failover-period={{ .Values.controllerManager.pdFailoverPeriod }} |
tennix
reviewed
Sep 3, 2018
| then | ||
| ARGS="${ARGS}--initial-cluster=${HOSTNAME}=http://${HOSTNAME}.${PEER_SERVICE_NAME}.${NAMESPACE}.svc:2380" | ||
| else | ||
| if [[ ${ORDINAL} -eq 0 ]] | ||
| then | ||
| TOP=$((replicas-1)) |
There was a problem hiding this comment.
replicas is undefined, define it like following:
replicas=${replicas:-3}
|
/run-e2e-tests |
|
/run-e2e-tests |
|
/run-e2e-tests |
tennix
reviewed
Sep 6, 2018
| if err != nil && !errors.IsNotFound(err) { | ||
| return err | ||
| } | ||
| // start all the pd members together, if this is a old cluster |
|
/run-e2e-tests |
|
/run-e2e-tests |
tennix
approved these changes
Sep 7, 2018
xiaojingchen
approved these changes
Sep 7, 2018
queenliuxx
pushed a commit
to queenliuxx/tidb-operator
that referenced
this pull request
Dec 19, 2018
* *: pd failover
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This PR fixes #57, adds PD failover feature to TiDB Operator.
@tennix @onlymellb @xiaojingchen @gregwebs PTAL