values-production.yaml

## @section Production Configuration for Memcached
##
## This is a production-ready configuration file with optimized settings
##
## Key Production Features:
## - High availability with multiple replicas
## - Hard pod anti-affinity for better fault tolerance
## - Enhanced resource limits and requests
## - Horizontal Pod Autoscaling enabled
## - Prometheus metrics and ServiceMonitor enabled
## - Topology spread constraints for zone/node distribution
## - Persistence enabled for extended storage (extstore)
##

## @param replicaCount Number of Memcached replicas for high availability
## Increase this for better availability and load distribution
## Recommended: 3-5 for production, 10-15+ for large scale deployments
replicaCount: 3

## @param podAntiAffinityPreset Pod anti-affinity preset to spread pods across nodes
## Values: soft (preferred) or hard (required)
## Hard anti-affinity ensures pods are never co-located on the same node
podAntiAffinityPreset: hard

## Resource Configuration
## Define resource requests and limits for production workloads
## Ref: https://kubernetes.io/docs/concepts/configuration/manage-resources-containers/
resources:
  ## @param resources.limits Resource limits for Memcached container
  ## Limits define the maximum resources a container can use
  limits:
    ## @param resources.limits.memory Maximum memory allocation
    ## Should be higher than requests to allow for burst capacity
    memory: 2Gi
    ## @param resources.limits.cpu Maximum CPU allocation in millicores
    ## 1000m = 1 CPU core
    cpu: 1000m
  ## @param resources.requests Resource requests for Memcached container
  ## Requests define the guaranteed resources for scheduling
  requests:
    ## @param resources.requests.memory Guaranteed memory allocation
    ## Set based on expected baseline usage
    memory: 1Gi
    ## @param resources.requests.cpu Guaranteed CPU allocation in millicores
    ## 500m = 0.5 CPU cores
    cpu: 500m

## @section Memcached Configuration
## Enhanced production configuration
memcached:
  ## @param memcached.allocatedMemory Memory allocated for memcached cache in megabytes
  ## This is the main cache memory (in-memory storage)
  ## Adjust based on your workload requirements:
  ## - Small deployments: 256-512 MB
  ## - Medium deployments: 512-1024 MB
  ## - Large deployments: 1024-4096 MB
  ## - Very large deployments: 4096+ MB
  ## Note: Should be less than container memory limit
  allocatedMemory: 1536

  ## @param memcached.maxItemMemory Maximum size per item in megabytes
  ## Default is 1 MB, increase for larger objects
  ## Common values:
  ## - 1 MB: Default, suitable for most use cases
  ## - 5-10 MB: Medium-sized objects
  ## - 10-50 MB: Large objects (images, documents)
  maxItemMemory: 10

  ## @param memcached.connectionLimit Maximum concurrent connections
  ## Increase for high-concurrency production environments
  ## Guidelines:
  ## - Small: 1024 connections
  ## - Medium: 2048 connections
  ## - Large: 4096+ connections
  connectionLimit: 2048

  ## @param memcached.extendedOptions Extended memory and performance options
  ## Available options:
  ## - modern: Uses modern protocol defaults and performance optimizations
  ## - track_sizes: Tracks per-command statistics (useful for monitoring/debugging)
  ## - maxconns_fast: Close connections after accepting them for immediate reuse
  ## - lru_crawler: Enable LRU crawler for better memory management
  ## - lru_maintainer: Background thread for LRU maintenance
  extendedOptions: "modern,track_sizes"

  ## @param memcached.extraExtendedOptions Additional extended options for external storage
  ## Additional options to append to extendedOptions
  ## When persistence.enabled=true, extstore is automatically configured:
  ## - ext_path is set to {persistence.mountPath}/extstore:{calculated_size}G
  ## - ext_wbuf_size=16 is added automatically (can be overridden here)
  ## - calculated_size is 90% of persistence.size to allow filesystem overhead
  ##
  ## You can add other extstore options here:
  ## - ext_threads: Number of IO threads (default: 1, increase for high IOPS)
  ## - ext_wbuf_size: Override default write buffer (default: 16MB when auto-configured)
  ## - ext_item_size: Minimum item size to store in extstore (default: uses -I setting)
  ## - ext_item_age: Minimum age before moving to extstore (default: no age limit)
  ## - ext_max_sleep: Max time for IO threads to sleep (microseconds)
  ## - ext_recache_rate: Rate to recache items from extstore (items/sec)
  ##
  ## Example to increase IO threads:
  ## extraExtendedOptions: "ext_threads=4"
  ##
  ## NOTE: Do NOT specify ext_path here - it's automatically configured from persistence settings
  extraExtendedOptions: ""

  ## @param memcached.verbosity Logging verbosity level
  ## Options:
  ## - "" (empty): Quiet mode, minimal logging
  ## - "v": Verbose, logs important events (recommended for production)
  ## - "vv": Very verbose, detailed logging
  ## - "vvv": Extremely verbose, debug-level logging
  verbosity: "v"

  ## @param memcached.port Memcached listening port
  ## Default: 11211 (standard Memcached port)
  port: 11211

## @section Persistence Configuration
## Enable persistent storage for Memcached extstore feature
## Extstore allows Memcached to extend cache to disk for larger datasets
## Ref: https://github.com/memcached/memcached/wiki/Extstore
persistence:
  ## @param persistence.enabled Enable persistence using Persistent Volume Claims
  ## MUST be enabled when using memcached.extraExtendedOptions with ext_path
  ## Provides disk-backed storage for extended cache capacity
  enabled: true
  ## @param persistence.storageClass Persistent Volume storage class
  ## Leave empty to use the default storage class
  ## Examples: "standard", "fast-ssd", "ebs-sc", "ceph-rbd"
  storageClass: ""
  ## @param persistence.size Persistent Volume size
  ## Must match or exceed the size specified in extraExtendedOptions (50G)
  ## Recommended: Add 10-20% overhead for filesystem metadata
  ## Examples: 10Gi, 50Gi, 100Gi, 500Gi
  size: 50Gi
  ## @param persistence.mountPath Path where the volume will be mounted
  ## Must match the path specified in extraExtendedOptions ext_path
  ## Default: /data (used in ext_path=/data/extstore:50G)
  mountPath: /data
  ## @param persistence.accessModes Persistent Volume access modes
  ## ReadWriteOnce: Volume can be mounted as read-write by a single node
  ## Suitable for StatefulSet with one pod per volume
  accessModes:
    - ReadWriteOnce

## @section Extra Arguments Configuration
## @param extraArgs Additional command-line arguments for Memcached
## Use this to pass additional flags to the memcached binary
## Format: Array of key-value pairs
## Common production arguments:
## - idle_timeout: Close connections idle for this many seconds (default: 0 = disabled)
## - no_relisten: Disable automatic rebinding after network errors
## - disable_cas: Disable CAS (Compare-And-Swap) command
## - disable_flush: Disable flush_all command for security
##
## Examples:
extraArgs: []
## Uncomment to enable idle timeout (15 minutes):
# extraArgs:
#   - key: "idle_timeout"
#     value: "900"
## Uncomment to disable automatic rebinding:
#   - key: "no_relisten"
#     value: ""
## Uncomment to disable flush_all command for security:
#   - key: "disable_flush"
#     value: ""

## @section Metrics and Monitoring Configuration
## Prometheus metrics exporter configuration
metrics:
  ## @param metrics.enabled Enable Prometheus metrics exporter sidecar
  ## Deploys prom/memcached-exporter as a sidecar container
  enabled: true
  ## Metrics exporter image configuration
  image:
    ## @param metrics.image.repository Memcached exporter image repository
    repository: prom/memcached-exporter
    ## @param metrics.image.tag Memcached exporter image tag
    ## Check https://github.com/prometheus/memcached_exporter/releases for latest
    tag: v0.14.4
  ## @param metrics.containerPort Port where metrics are exposed
  ## Default Prometheus exporter port
  containerPort: 9150
  ## @param metrics.resources Resource limits and requests for metrics exporter
  ## Keep this lightweight as it's just for metrics collection
  resources:
    ## @param metrics.resources.limits Maximum resources for exporter
    limits:
      memory: 256Mi
      cpu: 200m
    ## @param metrics.resources.requests Guaranteed resources for exporter
    requests:
      memory: 128Mi
      cpu: 100m
  ## ServiceMonitor configuration for Prometheus Operator
  ## Ref: https://github.com/prometheus-operator/prometheus-operator
  serviceMonitor:
    ## @param metrics.serviceMonitor.enabled Create ServiceMonitor resource
    ## Requires Prometheus Operator to be installed
    enabled: true
    ## @param metrics.serviceMonitor.interval How often to scrape metrics
    ## Common values: 15s, 30s, 60s
    interval: 30s
    ## @param metrics.serviceMonitor.additionalLabels Labels to help Prometheus discover this ServiceMonitor
    ## Must match your Prometheus serviceMonitorSelector configuration
    additionalLabels:
      prometheus: kube-prometheus

## @section Autoscaling Configuration
## Horizontal Pod Autoscaler (HPA) for dynamic scaling
## Ref: https://kubernetes.io/docs/tasks/run-application/horizontal-pod-autoscale/
autoscaling:
  ## @param autoscaling.enabled Enable Horizontal Pod Autoscaler
  ## Automatically scales replicas based on CPU and memory usage
  enabled: true
  ## @param autoscaling.minReplicas Minimum number of replicas
  ## Should match or be close to replicaCount for production stability
  minReplicas: 3
  ## @param autoscaling.maxReplicas Maximum number of replicas
  ## Set based on cluster capacity and expected maximum load
  ## Common values: 10-20 for medium scale, 50+ for large scale
  maxReplicas: 15
  ## @param autoscaling.targetCPU Target CPU utilization percentage
  ## Scale up when average CPU exceeds this threshold
  ## 70-80% is recommended for production to handle spikes
  targetCPU: 70
  ## @param autoscaling.targetMemory Target Memory utilization percentage
  ## Scale up when average memory exceeds this threshold
  ## 80% allows for some burst capacity
  targetMemory: 80

## @section Topology and Distribution Configuration
## @param topologySpreadConstraints Topology spread constraints for better pod distribution
## Ensures pods are distributed across zones and nodes for high availability
## Ref: https://kubernetes.io/docs/concepts/scheduling-eviction/topology-spread-constraints/
##
## This configuration ensures:
## 1. Pods are spread across different availability zones (zone constraint)
## 2. Pods are spread across different nodes (hostname constraint)
## 3. DoNotSchedule ensures pods won't be scheduled if constraints can't be met
topologySpreadConstraints:
  ## Zone-level distribution constraint
  ## @param topologySpreadConstraints[0].maxSkew Maximum allowed difference in pod count between zones
  ## maxSkew: 1 means zones should have equal or near-equal number of pods
  - maxSkew: 1
    ## @param topologySpreadConstraints[0].topologyKey Topology key for zone distribution
    topologyKey: topology.kubernetes.io/zone
    ## @param topologySpreadConstraints[0].whenUnsatisfiable Action when constraint cannot be satisfied
    ## DoNotSchedule: Don't schedule the pod (strict enforcement)
    ## ScheduleAnyway: Try to meet constraint but schedule anyway (soft enforcement)
    whenUnsatisfiable: DoNotSchedule
    ## @param topologySpreadConstraints[0].labelSelector Label selector to identify related pods
    labelSelector:
      matchLabels:
        app.kubernetes.io/name: memcached
  ## Node-level distribution constraint
  ## @param topologySpreadConstraints[1].maxSkew Maximum allowed difference in pod count between nodes
  - maxSkew: 1
    ## @param topologySpreadConstraints[1].topologyKey Topology key for node distribution
    topologyKey: kubernetes.io/hostname
    ## @param topologySpreadConstraints[1].whenUnsatisfiable Action when constraint cannot be satisfied
    whenUnsatisfiable: DoNotSchedule
    ## @param topologySpreadConstraints[1].labelSelector Label selector to identify related pods
    labelSelector:
      matchLabels:
        app.kubernetes.io/name: memcached

## @section Pod Labels and Annotations
## @param podLabels Additional labels for Memcached pods
## Useful for monitoring, organization, and service discovery
## These labels help with:
## - Identifying environment in monitoring dashboards
## - Filtering logs and metrics
## - Service mesh integration
## - Cost tracking and resource attribution
podLabels:
  environment: production
  tier: cache

## @section Health Check Configuration
## @param livenessProbe Liveness probe configuration
## Kubernetes restarts the pod if this probe fails
## Ref: https://kubernetes.io/docs/tasks/configure-pod-container/configure-liveness-readiness-startup-probes/
livenessProbe:
  ## @param livenessProbe.enabled Enable liveness probe
  enabled: true
  ## @param livenessProbe.initialDelaySeconds Initial delay before probe starts
  ## Give Memcached time to start up
  initialDelaySeconds: 30
  ## @param livenessProbe.periodSeconds How often to perform the probe
  periodSeconds: 10
  ## @param livenessProbe.timeoutSeconds Timeout for the probe
  timeoutSeconds: 5
  ## @param livenessProbe.failureThreshold Number of failures before restarting
  ## 6 failures = 60 seconds before restart (with 10s period)
  failureThreshold: 6

## @param readinessProbe Readiness probe configuration
## Kubernetes removes pod from service endpoints if this probe fails
readinessProbe:
  ## @param readinessProbe.enabled Enable readiness probe
  enabled: true
  ## @param readinessProbe.initialDelaySeconds Initial delay before probe starts
  ## Shorter than liveness as readiness checks if cache is ready to serve
  initialDelaySeconds: 5
  ## @param readinessProbe.periodSeconds How often to perform the probe
  periodSeconds: 5
  ## @param readinessProbe.timeoutSeconds Timeout for the probe
  timeoutSeconds: 3
  ## @param readinessProbe.failureThreshold Number of failures before marking unready
  ## 3 failures = 15 seconds before removing from endpoints
  failureThreshold: 3

## @section Service Configuration
## @param service Kubernetes Service configuration
service:
  ## @param service.type Service type (ClusterIP for internal access)
  type: ClusterIP
  ## @param service.annotations Service annotations
  ## topology-aware-hints enables topology-aware routing for better latency
  ## Ref: https://kubernetes.io/docs/concepts/services-networking/topology-aware-hints/
  annotations:
    service.kubernetes.io/topology-aware-hints: auto

## @section Pod Management Configuration
## @param terminationGracePeriodSeconds Grace period for pod termination
## Allows time for connections to drain and cache to flush
## 60 seconds is recommended for production to ensure clean shutdown
terminationGracePeriodSeconds: 60

## @param podManagementPolicy StatefulSet pod management policy
## Parallel: Start/stop all pods simultaneously (faster scaling)
## OrderedReady: Start/stop pods in order (0, 1, 2...)
podManagementPolicy: Parallel