diff --git a/charts/redis/Chart.yaml b/charts/redis/Chart.yaml
index 8d0060aae313562bfd81b3ad7fe2d28a4d9f1dae..3359b925d4e42d7bc6daab7adfbc6a06969d75db 100644
--- a/charts/redis/Chart.yaml
+++ b/charts/redis/Chart.yaml
@@ -8,7 +8,7 @@ maintainers:
- name: groundhog2k
# This is the chart version
-version: 0.5.10
+version: 0.6.0
# This is the version number of the application being deployed.
appVersion: "7.0.7"
diff --git a/charts/redis/README.md b/charts/redis/README.md
index 73d286a54962e72b300255ea9e047521434b7fe5..d36e9393a91b3fc4a01ea8cfa97742305f68beb7 100644
--- a/charts/redis/README.md
+++ b/charts/redis/README.md
@@ -1,6 +1,6 @@
# Redis
-  
+  
## Changelog
@@ -160,11 +160,13 @@ helm uninstall my-release
| extraSentinelSecrets[].name | string | `nil` | Name of the existing K8s secret |
| extraSentinelSecrets[].mountPath | string | `nil` | Mount path where the secret should be mounted into the container (f.e. /mysecretfolder) |
| haMode.enabled | bool | `false` | Enable Redis high availibility mode with master-slave replication and sentinel |
+| haMode.useDnsNames | bool | `false` | Use DNS names instead of Pod IPs to build the cluster |
| haMode.masterGroupName | string | `"redisha"` | Mandatory redis HA-master group name |
| haMode.replicas | int | `3` | Number of replicas (minimum should be 3) |
| haMode.quorum | int | `2` | Quorum of sentinels that need to agree that a master node is not available |
-| haMode.downAfterMilliseconds | int | `5000` | Number of milliseconds after the master should be declared as unavailable |
-| haMode.failoverTimeout | int | `180000` | Timeout for a failover |
-| haMode.parallelSyncs | int | `2` | Number of parallel reconfigurations
+| haMode.downAfterMilliseconds | int | `30000` | Number of milliseconds after the master should be declared as unavailable |
+| haMode.failoverTimeout | int | `180000` | Timeout for a failover in milliseoncds |
+| haMode.parallelSyncs | int | `1` | Number of parallel reconfigurations
| haMode.masterAliveTestTimeout | int | `2` | Timeout in seconds to detect if Redis master is alive |
-| haMode.failoverWait | int | `5` | Assumed wait time in seconds until failover should be finished |
+| haMode.failoverWait | int | `35` | Assumed wait time in seconds until failover should be finished and before failover will be forced (should be greater than value of downAfterMilliseconds) |
+| haMode.keepOldLogs | bool | `false` | Keep old init logs in /data/init.log after a successful initialization (use only for debugging) |
diff --git a/charts/redis/RELEASENOTES.md b/charts/redis/RELEASENOTES.md
index f9ed6b874364e5439a0e5eb6dd22c8a0d5acba61..173add1da9ace1328335395a23c010dd80bc5ea6 100644
--- a/charts/redis/RELEASENOTES.md
+++ b/charts/redis/RELEASENOTES.md
@@ -16,4 +16,5 @@
| 0.5.8 | 7.0.5 | Implemented support for image.registry option |
| 0.5.9 | 7.0.6 | Upgraded to Redis 7.0.6 and added support for init container resources |
| 0.5.10 | 7.0.7 | Upgraded to Redis 7.0.7 |
+| 0.6.0 | 7.0.7 | Implemented alternative DNS option for cluster building |
| | | |
diff --git a/charts/redis/templates/scripts.yaml b/charts/redis/templates/scripts.yaml
index f52cf070358de31efafc8eb2d49bcb26d10de26a..e52ec19ffbd163938e005a1eed9969e6634fcb18 100644
--- a/charts/redis/templates/scripts.yaml
+++ b/charts/redis/templates/scripts.yaml
@@ -6,98 +6,167 @@ metadata:
{{- include "redis.labels" . | nindent 4 }}
data:
init.sh: |
- #!/bin/sh
- {{- if .Values.haMode.enabled }}
- MASTER="$(timeout {{ .Values.haMode.masterAliveTestTimeout }}s redis-cli -h {{ template "redis.fullname" . }} -p {{ .Values.service.sentinelPort }} sentinel get-master-addr-by-name {{ include "redis.masterGroupName" . }} | grep -E '[0-9]{1,3}\.[0-9]{1,3}\.[0-9]{1,3}\.[0-9]{1,3}')"
- {{- end }}
+ #!/bin/bash
REDIS_CONFIG=/data/conf/redis.conf
SENTINEL_CONFIG=/data/conf/sentinel.conf
+ if [ -f "/data/init.log" ]; then
+ echo "Detected restart of this instance ($HOSTNAME)"
+ echo "## This was the previous log:"
+ cat /data/init.log
+ echo "## End of previous log"
+ fi
- {{- if .Values.haMode.enabled }}
- test_master_alive() {
- TEST="$(timeout {{ .Values.haMode.masterAliveTestTimeout }}s redis-cli -h $MASTER -p 6379 ping)"
- if [ -z "$TEST" ]; then
- return 1
- fi
- return 0
+ # Log a message during initialization phase
+ # $1 - the log message
+ log() {
+ echo "$(date) $1"
+ echo "$(date) $1" >>/data/init.log
}
- {{- end }}
- configure_redis() {
- echo "Configuring redis server..."
- rm -f $REDIS_CONFIG
+ # Creating redis base configuration
+ configure_redis_base() {
+ log "Creating redis base configuration"
+ mkdir -p /data/conf
+ rm -f $REDIS_CONFIG
- echo "Setting redis server defaults"
- echo "port 6379" >>$REDIS_CONFIG
- echo "protected-mode no" >>$REDIS_CONFIG
- echo "bind 0.0.0.0" >>$REDIS_CONFIG
- echo "dir /data" >>$REDIS_CONFIG
+ log "Setting redis server defaults"
+ echo "port 6379" >>$REDIS_CONFIG
+ echo "protected-mode no" >>$REDIS_CONFIG
+ echo "bind 0.0.0.0" >>$REDIS_CONFIG
+ echo "dir /data" >>$REDIS_CONFIG
+ log "Finished creating base configuration"
+ }
- {{- if .Values.haMode.enabled }}
- if [ -z "$MASTER" ]; then
- echo "No master found - This instance ($HOSTNAME) will be master now"
- else
- echo "Redis master was found with address $MASTER - Checking if master is still alive "
- test_master_alive
- if [ "$?" -eq "1" ]; then
- echo "Dead master at address $MASTER detected! - waiting for failover"
- sleep {{ .Values.haMode.failoverWait }}
- test_master_alive
- if [ "$?" -eq "1" ]; then
- echo "Master is still dead! - forcing failover and retry pod initialization"
- redis-cli -h {{ template "redis.fullname" . }} -p {{ .Values.service.sentinelPort }} sentinel failover {{ include "redis.masterGroupName" . }}
- exit 1
- fi
- fi
- echo "Setting this instance ($HOSTNAME) as replicaof $MASTER"
- echo "replicaof $MASTER 6379" >>$REDIS_CONFIG
- fi
- {{- end }}
+ configure_redis_ext() {
+ if [ -f /usr/local/etc/redis/redis.conf ]; then
+ log "Adding optional redis configuration"
+ cat /usr/local/etc/redis/redis.conf >>$REDIS_CONFIG
+ fi
+ if [ -d /extraredisconfigs ]; then
+ log "Adding extra redis configs to redis configuration"
+ cat /extraredisconfigs/* >>$REDIS_CONFIG
+ fi
+ }
- if [ -f /usr/local/etc/redis/redis.conf ]; then
- echo "Adding optional redis configuration settings"
- cat /usr/local/etc/redis/redis.conf >>$REDIS_CONFIG
- fi
- if [ -d /extraredisconfigs ]; then
- echo "Adding extra redis configs to redis configuration"
- cat /extraredisconfigs/* >>$REDIS_CONFIG
- fi
- echo "Configuring redis server finished."
+ test_master_alive() {
+ TEST="$(timeout {{ .Values.haMode.masterAliveTestTimeout }}s redis-cli -h $MASTER -p 6379 ping)"
+ if [ -z "$TEST" ]; then
+ log "Master is not alive"
+ return 1
+ fi
+ log "Master is alive"
+ return 0
}
- {{- if .Values.haMode.enabled }}
+
+ test_valid_dns() {
+ KNOWN_HOSTS=($(getent hosts {{ include "redis.fullname" . }}-headless.{{ .Release.Namespace }}.svc.{{ .Values.clusterDomain }} | awk '{ print $1 }'))
+ log "## Known hosts for headless service {{ include "redis.fullname" . }}-headless.{{ .Release.Namespace }}.svc.{{ .Values.clusterDomain }}:"
+ for AHOSTIP in "${KNOWN_HOSTS[@]}"; do
+ AHOSTNAME=$(getent hosts $AHOSTIP | awk '{print $2}')
+ log "${AHOSTIP} ${AHOSTNAME}"
+ done
+ log "#####"
+ {{- if .Values.haMode.useDnsNames }}
+ MASTERENTRY="$(getent hosts $MASTER | awk '{ print $2 }')"
+ {{- else }}
+ MASTERENTRY="$(getent hosts $MASTER | awk '{ print $1 }')"
+ {{- end }}
+ }
+
configure_sentinel() {
- echo "Redis HA-mode enabled"
- echo "Configuring sentinel server..."
- rm -f $SENTINEL_CONFIG
+ log "Configuring sentinel server..."
+ rm -f $SENTINEL_CONFIG
+
+ log "Setting sentinel defaults"
+ if [ -z "$MASTER" ]; then
+ {{- if .Values.haMode.useDnsNames }}
+ MASTER=${HOSTNAME}.{{ include "redis.fullname" . }}-headless.{{ .Release.Namespace }}.svc.{{ .Values.clusterDomain }}
+ {{- else }}
+ MASTER="$(getent hosts $HOSTNAME | awk '{ print $1 }')"
+ {{- end }}
+ log "No master found - Configuring sentinel for master $HOSTNAME with address $MASTER"
+ echo "sentinel monitor {{ include "redis.masterGroupName" . }} $MASTER 6379 {{ .Values.haMode.quorum }}" >>$SENTINEL_CONFIG
+ else
+ log "Redis master was found - Configuring sentinel for master address $MASTER"
+ echo "sentinel monitor {{ include "redis.masterGroupName" . }} $MASTER 6379 {{ .Values.haMode.quorum }}" >>$SENTINEL_CONFIG
+ fi
+ {{- if .Values.haMode.useDnsNames }}
+ echo "SENTINEL resolve-hostnames yes" >>$SENTINEL_CONFIG
+ echo "SENTINEL announce-hostnames yes" >>$SENTINEL_CONFIG
+ {{- end }}
+ echo "sentinel down-after-milliseconds {{ include "redis.masterGroupName" . }} {{ .Values.haMode.downAfterMilliseconds }}" >>$SENTINEL_CONFIG
+ echo "sentinel failover-timeout {{ include "redis.masterGroupName" . }} {{ .Values.haMode.failoverTimeout }}" >>$SENTINEL_CONFIG
+ }
+
+ configure_sentinel_ext() {
+ if [ -f /usr/local/etc/redis/sentinel.conf ]; then
+ log "Adding optional sentinel configuration settings"
+ cat /usr/local/etc/redis/sentinel.conf >>$SENTINEL_CONFIG
+ fi
+ if [ -d /extrasentinelconfigs ]; then
+ log "Adding extra sentinel configs to sentinel configuration"
+ cat /extrasentinelconfigs/* >>$SENTINEL_CONFIG
+ fi
+ log "Configuring sentinel server finished."
+ }
- echo "Setting sentinel defaults"
- if [ -z "$MASTER" ]; then
- MASTER_IP="$(getent hosts $HOSTNAME | awk '{ print $1 }')"
- echo "No master found - Configuring sentinel for master $HOSTNAME with address $MASTER_IP"
- echo "sentinel monitor {{ include "redis.masterGroupName" . }} $MASTER_IP 6379 {{ .Values.haMode.quorum }}" >>$SENTINEL_CONFIG
+ configure_redis_cluster() {
+ log "Try to detect cluster master.."
+ MASTER="$(timeout {{ .Values.haMode.masterAliveTestTimeout }}s redis-cli -h {{ template "redis.fullname" . }} -p {{ .Values.service.sentinelPort }} sentinel get-master-addr-by-name {{ include "redis.masterGroupName" . }} | head -n 1)"
+ if [ -z "$MASTER" ]; then
+ log "No master found - This instance ($HOSTNAME) will be master now"
+ else
+ log "Redis master was found with address $MASTER - Checking host dns entry"
+ test_valid_dns
+ if [ -z "$MASTERENTRY" ]; then
+ log "No valid DNS entry found!"
+ if [ -f "/data/failover_restart" ]; then
+ rm /data/failover_restart
+ log "Forcing failover now"
+ redis-cli -h {{ template "redis.fullname" . }} -p {{ .Values.service.sentinelPort }} sentinel failover {{ include "redis.masterGroupName" . }}
+ else
+ log "Waiting for failover before restart"
+ sleep {{ .Values.haMode.failoverWait }}
+ touch /data/failover_restart
+ fi
+ log "Restart"
+ exit 1
else
- echo "Redis master was found - Configuring sentinel for master address $MASTER"
- echo "sentinel monitor {{ include "redis.masterGroupName" . }} $MASTER 6379 {{ .Values.haMode.quorum }}" >>$SENTINEL_CONFIG
- fi
- echo "sentinel down-after-milliseconds {{ include "redis.masterGroupName" . }} {{ .Values.haMode.downAfterMilliseconds }}" >>$SENTINEL_CONFIG
- echo "sentinel failover-timeout {{ include "redis.masterGroupName" . }} {{ .Values.haMode.failoverTimeout }}" >>$SENTINEL_CONFIG
-
- if [ -f /usr/local/etc/redis/sentinel.conf ]; then
- echo "Adding optional sentinel configuration settings"
- cat /usr/local/etc/redis/sentinel.conf >>$SENTINEL_CONFIG
- fi
- if [ -d /extrasentinelconfigs ]; then
- echo "Adding extra sentinel configs to sentinel configuration"
- cat /extrasentinelconfigs/* >>$SENTINEL_CONFIG
+ MASTER=$MASTERENTRY
+ log "$MASTER has valid DNS entry"
+ log "Checking if master is alive"
+ test_master_alive
+ if [ "$?" -eq "1" ]; then
+ if [ -f "/data/failover_restart" ]; then
+ rm /data/failover_restart
+ log "Master is still dead! - forcing failover and retry pod initialization"
+ redis-cli -h {{ template "redis.fullname" . }} -p {{ .Values.service.sentinelPort }} sentinel failover {{ include "redis.masterGroupName" . }}
+ else
+ log "Dead master at address $MASTER detected! - waiting for failover"
+ sleep {{ .Values.haMode.failoverWait }}
+ touch /data/failover_restart
+ fi
+ log "Restart"
+ exit 1
+ else
+ rm -f /data/failover_restart
+ log "Setting this instance ($HOSTNAME) as replicaof $MASTER"
+ echo "replicaof $MASTER 6379" >>$REDIS_CONFIG
+ fi
fi
- echo "Configuring sentinel server finished."
+ fi
}
- {{- end }}
- echo "Initialize configuration..."
- mkdir -p /data/conf
- configure_redis
+ log "Creating configuration..."
+ configure_redis_base
{{- if .Values.haMode.enabled }}
+ log "Redis HA-mode is enabled"
+ configure_redis_cluster
configure_sentinel
+ configure_sentinel_ext
+ {{- end }}
+ configure_redis_ext
+ log "Done."
+ {{- if not .Values.haMode.keepOldLogs }}
+ rm -f /data/init.log
{{- end }}
- echo "Finished."
diff --git a/charts/redis/values.yaml b/charts/redis/values.yaml
index abda3f0ead2caa42a9d92fec44efaf8c481c0021..cc3839addaf8ac9802add794bfe9b2a3c5f47dfd 100644
--- a/charts/redis/values.yaml
+++ b/charts/redis/values.yaml
@@ -151,6 +151,9 @@ extraInitContainers: []
## Extra containers for usage as sidecars
extraContainers: []
+## Default Kubernetes cluster domain
+clusterDomain: cluster.local
+
## Arguments for the container entrypoint process (Redis server)
args: []
@@ -195,6 +198,8 @@ extraSentinelSecrets: []
haMode:
## Enable high availibility deployment mode
enabled: false
+ ## Use DNS names instead of Pod IPs to build the cluster
+ useDnsNames: false
## Mandatory redis HA-master group name (default "redisha")
masterGroupName: "redisha"
## Number of replicas (minimum should be 3)
@@ -202,15 +207,17 @@ haMode:
## Quorum of sentinels that need to agree that a master node is not available
quorum: 2
## Number of milliseconds after the master should be declared as unavailable
- downAfterMilliseconds: 5000
+ downAfterMilliseconds: 30000
## Timeout for a failover
failoverTimeout: 180000
## Number of parallel reconfigurations
- parallelSyncs: 2
+ parallelSyncs: 1
## Timeout in seconds to detect if Redis master is alive
masterAliveTestTimeout: 2
- ## Assumed wait time in seconds until failover should be finished
- failoverWait: 5
+ ## Assumed wait time in seconds until failover should be finished and before failover will be forced (should be greater than value of downAfterMilliseconds)
+ failoverWait: 35
+ ## Keep old init logs in /data/init.log after a successful initialization (use only for debugging)
+ keepOldLogs: false
## Storage parameters
storage: