add option to set ingesters podManagementPolicy (#236)

* add option to set ingesters podManagementPolicy

Signed-off-by: Andre Ziviani <[email protected]>

* update changelog

Signed-off-by: Andre Ziviani <[email protected]>

* update readme

Signed-off-by: Andre Ziviani <[email protected]>

* add a warning about podmanagementpolicy

Signed-off-by: Andre Ziviani <[email protected]>

Author: AndreZiviani

CHANGELOG.md

@@ -17,6 +17,7 @@
 * [ENHANCEMENT] Optionally generate endpoints for `X-Scope-OrgID` injection (multi-tenancy) #180
 * [ENHANCEMENT] Optionally configure Basic Auth within Nginx #205
 * [ENHANCEMENT] Add extraEndpointSpec to all servicemonitors for specifying things like TLS #216
+* [ENHANCEMENT] Add option to specify `podManagementPolicy` when using ingesters as statefulset #236
 * [BUGFIX] Fix whitespace in runtime-config annotations, introduced in #209, fixed in #212
 * [BUGFIX] Correcting nginx config for auth orgs to right proxy_pass #192
 * [DEPENDENCY] Update Helm release memcached to v5.15.4 #228

README.md

@@ -434,6 +434,7 @@ Kubernetes: `^1.19.0-0`
 | ingester.​startupProbe.​initialDelaySeconds | int | `120` |  |
 | ingester.​startupProbe.​periodSeconds | int | `30` |  |
 | ingester.​statefulSet.​enabled | bool | `false` |  |
+| ingester.​statefulSet.​podManagementPolicy | string | `"OrderedReady"` | see https://cortexmetrics.io/docs/guides/ingesters-scaling-up-and-down/#scaling-down and https://kubernetes.io/docs/concepts/workloads/controllers/statefulset/#pod-management-policies for scaledown details |
 | ingester.​statefulStrategy.​type | string | `"RollingUpdate"` |  |
 | ingester.​strategy.​rollingUpdate.​maxSurge | int | `0` |  |
 | ingester.​strategy.​rollingUpdate.​maxUnavailable | int | `1` |  |

templates/ingester/ingester-statefulset.yaml

@@ -18,6 +18,7 @@ spec:
       {{- include "cortex.ingesterSelectorLabels" . | nindent 6 }}
   updateStrategy:
     {{- toYaml .Values.ingester.statefulStrategy | nindent 4 }}          
+  podManagementPolicy: "{{ .Values.ingester.statefulSet.podManagementPolicy }}"
   serviceName: {{ template "cortex.fullname" . }}-ingester-headless
   {{- if .Values.ingester.persistentVolume.enabled }}
   volumeClaimTemplates:

values.yaml

@@ -507,6 +507,8 @@ ingester:
     ## This is useful when using WAL
     ##
     enabled: false
+    # -- see https://cortexmetrics.io/docs/guides/ingesters-scaling-up-and-down/#scaling-down and https://kubernetes.io/docs/concepts/workloads/controllers/statefulset/#pod-management-policies for scaledown details
+    podManagementPolicy: OrderedReady
 
   service:
     annotations: {}