Merge pull request #83 from Altinity/release-0.2.2

Release 0.2.2

.gitignore

@@ -16,3 +16,4 @@
 
 # Skip log folder
 log
+dev/log

README.md

@@ -13,11 +13,13 @@ ClickHouse Operator is in beta. You can use it at your own risk. There may be ba
 
 The ClickHouse Operator for Kubernetes currently provides the following:
 
-- Creates cluster of the ClickHouse database based on Custom Resource [specification][crd_spec] provided
-- Supports Storage customization (VolumeClaim templates)
-- Supports Pod template customization (Volume and Container templates)
-- Supports ClickHouse configuration customization (including Zookeeper integration)
-- Supports ClickHouse metrics export to Prometheus
+- Creates ClickHouse cluster based on Custom Resource [specification][crd_spec] provided
+- Storage customization (VolumeClaim templates)
+- Pod template customization (Volume and Container templates)
+- ClickHouse configuration customization (including Zookeeper integration)
+- ClickHouse cluster scaling including automatic schema propagation
+- ClickHouse cluster version upgrades
+- Exporting ClickHouse metrics to Prometheus
 
 ## Requirements
 
@@ -40,11 +42,14 @@ The ClickHouse Operator for Kubernetes currently provides the following:
  * Adding shards and replicas
    * [Automatic schema creation][schema_migration]
  * [Update ClickHouse version][update_clickhouse_version]
+ * [Update Operator version][update_operator]
 
 **Monitoring**
  * [Prometheus & clickhouse-operator integration][prometheus_setup]
  * [Grafana & Prometheus integration][grafana_setup]
 
+**All docs**
+ * [All available docs list][all_docs_list]
 ## License
 
 Copyright (c) 2019, Altinity Ltd and/or its affiliates. All rights reserved.
@@ -65,5 +70,7 @@ See [LICENSE](./LICENSE) for more details.
 [storage]: ./docs/storage.md
 [update_cluster_add_replication]: ./docs/chi_update_add_replication.md
 [update_clickhouse_version]: ./docs/chi_update_clickhouse_version.md
+[update_operator]: ./docs/operator_upgrade.md
 [schema_migration]: ./docs/schema_migration.md
 [operator_configuration]: ./docs/operator_configuration.md
+[all_docs_list]: ./docs/README.md

cmd/clickhouse-operator/app/clickhouse_operator.go

@@ -44,7 +44,7 @@ import (
 )
 
 // Version defines current build versionRequest
-const Version = "0.2.0-beta"
+const Version = "0.2.2"
 
 // Prometheus exporter defaults
 const (

dev/binary_build.sh

@@ -0,0 +1,11 @@
+#!/bin/bash
+
+# Build clickhouse-operator
+# Do not forget to update version
+
+source ./binary_build_config.sh
+
+#CGO_ENABLED=0 GOOS=linux GOARCH=amd64 go build -o ${CUR_DIR}/clickhouse-operator ${SRC_ROOT}/cmd/clickhouse-operator
+CGO_ENABLED=0 go build -o ${OPERATOR_BIN} ${SRC_ROOT}/cmd/clickhouse-operator
+
+exit $?

dev/binary_build_config.sh

@@ -0,0 +1,7 @@
+#!/bin/bash
+
+# Build configuration options
+
+CUR_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" >/dev/null 2>&1 && pwd)"
+SRC_ROOT=$(realpath ${CUR_DIR}/..)
+OPERATOR_BIN=${CUR_DIR}/clickhouse-operator

dev/binary_clean.sh

@@ -0,0 +1,8 @@
+#!/bin/bash
+
+# Delete clickhouse-operator
+# Do not forget to update version
+
+source ./binary_build_config.sh
+
+rm -f ${OPERATOR_BIN}

dev/image_build_altinity.sh

@@ -0,0 +1,15 @@
+#!/bin/bash
+
+# Production docker image builder
+
+# Externally configurable build-dependent options
+TAG="${TAG:-altinity/clickhouse-operator:dev}"
+DOCKERHUB_LOGIN="${DOCKERHUB_LOGIN:-altinitybuilds}"
+DOCKERHUB_PUBLISH="${DOCKERHUB_PUBLISH:-yes}"
+MINIKUBE="${MINIKUBE:-no}"
+
+TAG="${TAG}" \
+DOCKERHUB_LOGIN="${DOCKERHUB_LOGIN}" \
+DOCKERHUB_PUBLISH="${DOCKERHUB_PUBLISH}" \
+MINIKUBE="${MINIKUBE}" \
+./image_build_universal.sh

dev/image_build_dev.sh

@@ -0,0 +1,15 @@
+#!/bin/bash
+
+# Dev docker image builder
+
+# Externally configurable build-dependent options
+TAG="${TAG:-sunsingerus/clickhouse-operator:dev}"
+DOCKERHUB_LOGIN="${DOCKERHUB_LOGIN:-sunsingerus}"
+DOCKERHUB_PUBLISH="${DOCKERHUB_PUBLISH:-yes}"
+MINIKUBE="${MINIKUBE:-no}"
+
+TAG="${TAG}" \
+DOCKERHUB_LOGIN="${DOCKERHUB_LOGIN}" \
+DOCKERHUB_PUBLISH="${DOCKERHUB_PUBLISH}" \
+MINIKUBE="${MINIKUBE}" \
+./image_build_universal.sh

dev/image_build_universal.sh

@@ -0,0 +1,28 @@
+#!/bin/bash
+
+# Universal docker image builder
+
+# Externally configurable build-dependent options
+TAG="${TAG:-sunsingerus/clickhouse-operator:dev}"
+DOCKERHUB_LOGIN="${DOCKERHUB_LOGIN:-sunsingerus}"
+DOCKERHUB_PUBLISH="${DOCKERHUB_PUBLISH:-yes}"
+MINIKUBE="${MINIKUBE:-no}"
+
+# Source-dependent options
+CUR_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" >/dev/null 2>&1 && pwd)"
+SRC_ROOT="$(realpath ${CUR_DIR}/..)"
+DOCKERFILE_DIR="${SRC_ROOT}"
+DOCKERFILE="${DOCKERFILE_DIR}/Dockerfile"
+
+# Build image with Docker
+if [[ "${MINIKUBE}" == "yes" ]]; then
+    # We'd like to build for minikube
+    eval $(minikube docker-env)
+fi
+cat "${DOCKERFILE}" | envsubst | docker build -t "${TAG}" "${SRC_ROOT}"
+
+# Publish image
+if [[ "${DOCKERHUB_PUBLISH}" == "yes" ]]; then
+    docker login -u "${DOCKERHUB_LOGIN}"
+    docker push "${TAG}"
+fi

run_dev.sh → dev/run_dev.sh

@@ -1,12 +1,18 @@
 #!/bin/bash
 
+# Run clickhouse-operator
+# Do not forget to update version
+
+source ./binary_build_config.sh
+LOG_DIR=${CUR_DIR}/log
+
 echo -n "Building binary, please wait..."
-if ./build_binary.sh; then
+if ./binary_build.sh; then
     echo "successfully built clickhouse-operator. Starting"
 
-    mkdir -p log
-    rm -f log/clickhouse-operator.*.log.*
-    ./clickhouse-operator \
+    mkdir -p ${LOG_DIR}
+    rm -f ${LOG_DIR}/clickhouse-operator.*.log.*
+    ${OPERATOR_BIN} \
     	-alsologtostderr=true \
     	-log_dir=log \
     	-v=1
@@ -19,12 +25,12 @@ if ./build_binary.sh; then
 # -stderrthreshold=FATAL Log events at or above this severity are logged to standard	error as well as to files
 
     # And clean binary after run. It'll be rebuilt next time
-    ./clean_binary.sh
+    ./binary_clean.sh
 
     echo "======================"
     echo "=== Logs available ==="
     echo "======================"
-    ls log/*
+    ls ${LOG_DIR}/*
 else
     echo "unable to build clickhouse-operator"
 fi

docs/README.md

@@ -0,0 +1,19 @@
+# Table of Contents
+1. [architecture.md](./architecture.md) - architecture overview
+1. [chi_update_add_replication.md](./chi_update_add_replication.md) - how to add replication
+1. [chi_update_clickhouse_version.md](./chi_update_clickhouse_version.md) - how to update version
+1. [clickhouse_config_errors_handling.md](./clickhouse_config_errors_handling.md) - how operator handles ClickHouse's config errors
+1. [custom_resource_explained.md](./custom_resource_explained.md) - explain Custom Resource Definition in details
+1. [grafana_setup.md](./grafana_setup.md) - how to setup Grafana
+1. [introduction.md](./introduction.md) - general introduction
+1. [k8s_cluster_access.md](./k8s_cluster_access.md) - how to setup cluster access
+1. [operator_build_from_sources.md](./operator_build_from_sources.md) - how to build operator from sources
+1. [operator_configuration.md](./operator_configuration.md) - operator configuration in details
+1. [operator_installation_details.md](./operator_installation_details.md) - how to install operator in details
+1. [operator_upgrade.md](./operator_upgrade.md) - how to upgrade operator to the different version
+1. [prometheus_setup.md](./prometheus_setup.md) - how to setup Prometheus
+1. [quick-start.md](./quick-start.md) - quick start
+1. [replication_setup.md](./replication_setup.md) - how to setup replication
+1. [schema_migration.md](./schema_migration.md) - how operator migrates shena during cluster resize
+1. [storage.md](./storage.md) - storage explained
+1. [zookeeper_setup.md](./zookeeper_setup.md) - how to setup zookeeper

docs/clickhouse_config_errors_handling.md

@@ -0,0 +1,36 @@
+# ClickHouse Configuration Errors Handling
+
+When operator performs rolling update and create procedures, it monitors health of the updated/created StatefulSets.
+Operator polls StatefulSet status and wait for some configurable time for StatefulSet to report **Ready** status reached. 
+In case StatefulSet successfully reports **Ready** status, all is good and operator can move on to the next StatefulSet.
+However, in case ClickHouse configuration is incorrect or due to any other reason ClickHouse is unable to start, StatefulSet would not reach **Ready** status.
+In tis case, operator has to do something and decide what to do regarding the following questions:
+1. What to do with current failed StatefulSet?
+1. Should operator continue with rolling update?
+
+Regarding failed StatefulSet, operator can either do nothing and leave the situation to admin to solve or can try to rollback StatefulSet to previous state.
+In case of newly created StatefulSet, rollback means *to delete StatefulSet*
+This behavior is configured with the following operator configuration options:
+```yaml
+# What to do in case created StatefulSet is not in Ready after `statefulSetUpdateTimeout` seconds
+# Possible options:
+# 1. abort - do nothing, just break the process and wait for admin
+# 2. delete - delete newly created problematic StatefulSet
+onStatefulSetCreateFailureAction: delete
+
+# What to do in case updated StatefulSet is not in Ready after `statefulSetUpdateTimeout` seconds
+# Possible options:
+# 1. abort - do nothing, just break the process and wait for admin
+# 2. rollback - delete Pod and rollback StatefulSet to previous Generation.
+# Pod would be recreated by StatefulSet based on rollback-ed configuration
+onStatefulSetUpdateFailureAction: rollback
+``` 
+
+Regarding should operator continue with rolling update/create in case of failed StatefulSet it met - current behavior is to abort rolling process and let admin to decide how to proceed with current situation.
+
+# Plans and discussion
+Interesting question is what to do with StatefulSets that were already successfully updated on the same run, before failed StatefulSet met.
+Available options are:
+1. Do nothing. In this case ClickHouse cluster may be in some inconsistent state, because some replicas may be updated and some not.
+1. try to rollback the whole cluster to some **previous** state. What this **previous** state be is a matter of discussion.
+Currently operator goes with 'do nothing' approach.

docs/examples-withstand-errors/01-incorrect-image-create.yaml

@@ -0,0 +1,58 @@
+apiVersion: "clickhouse.altinity.com/v1"
+kind: "ClickHouseInstallation"
+metadata:
+  name: "incorrect-image-create"
+spec:
+  configuration:
+    clusters:
+    - name: adv
+      layout:
+        type: Advanced
+        shards:
+        - definitionType: Replicas
+          replicas:
+          - deployment:
+              podTemplate: clickhouse:19.3.5
+          - deployment:
+              podTemplate: clickhouse:19.3.100
+          - deployment:
+              podTemplate: clickhouse:19.3.7
+
+  templates:
+    podTemplates:
+      - name: clickhouse:19.3.5
+        containers:
+          - name: clickhouse-pod
+            image: yandex/clickhouse-server:19.3.5
+            ports:
+              - name: http
+                containerPort: 8123
+              - name: client
+                containerPort: 9000
+              - name: interserver
+                containerPort: 9009
+
+      - name: clickhouse:19.3.7
+        containers:
+        - name: clickhouse-pod
+          image: yandex/clickhouse-server:19.3.7
+          ports:
+          - name: http
+            containerPort: 8123
+          - name: client
+            containerPort: 9000
+          - name: interserver
+            containerPort: 9009
+
+      # Incorrect image
+      - name: clickhouse:19.3.100
+        containers:
+        - name: clickhouse-pod
+          image: yandex/clickhouse-server:19.3.100
+          ports:
+          - name: http
+            containerPort: 8123
+          - name: client
+            containerPort: 9000
+          - name: interserver
+            containerPort: 9009

docs/examples-withstand-errors/02-incorrect-image-update-01-initial-position.yaml

@@ -0,0 +1,32 @@
+apiVersion: "clickhouse.altinity.com/v1"
+kind: "ClickHouseInstallation"
+metadata:
+  name: "incorrect-image-version-update"
+spec:
+  configuration:
+    clusters:
+      - name: update
+        deployment:
+          podTemplate: clickhouse:19.1.10
+        layout:
+          type: Advanced
+          shards:
+            - definitionType: Replicas
+              replicas:
+                - port: 9000
+                - port: 9000
+                - port: 9000
+
+  templates:
+    podTemplates:
+    - name: clickhouse:19.1.10
+      containers:
+      - name: clickhouse-pod
+        image: yandex/clickhouse-server:19.1.10
+        ports:
+        - name: http
+          containerPort: 8123
+        - name: client
+          containerPort: 9000
+        - name: interserver
+          containerPort: 9009