Support Accumulo installs on Microsoft Azure

* Add new cluster type 'azure' which leverages VM Scale Sets * Add HA (high-availability) capabilities for the Hadoop Name Node, Accumulo master, and Zookeeper roles within Muchos. Note: HA is on by default and should not be disabled * Enable central collection of metrics and logs using Azure Monitor * Increase some CentOS defaults to improve cluster stability * Fix latent bugs which prevent Spark from being set up correctly * Add checksums for specific Spark and Hadoop versions, as well as for Accumulo 2.0.0
apache · Aug 14, 2019 · be5ae7a · be5ae7a
1 parent 80281f2
commit be5ae7a
Show file tree

Hide file tree

Showing 61 changed files with 1,457 additions and 114 deletions.
diff --git a/README.md b/README.md
@@ -4,7 +4,7 @@
 
 **Muchos automates setting up [Apache Accumulo][accumulo] or [Apache Fluo][fluo] (and their dependencies) on a cluster**
 
-Muchos makes it easy to launch a cluster in Amazon's EC2 and deploy Accumulo or Fluo to it. Muchos
+Muchos makes it easy to launch a cluster in Amazon's EC2 or Microsoft Azure and deploy Accumulo or Fluo to it. Muchos
 enables developers to experiment with Accumulo or Fluo in a realistic, distributed environment.
 Muchos installs all software using tarball distributions which makes its easy to experiment
 with the latest versions of Accumulo, Hadoop, Zookeeper, etc without waiting for downstream packaging.
@@ -17,35 +17,67 @@ Muchos is structured into two high level components:
 
  * [Ansible] scripts that install and configure Fluo and its dependencies on a cluster.
  * Python scripts that push the Ansible scripts from a local development machine to a cluster and
- run them. These Python scripts can also optionally launch a cluster in EC2 using [boto].
+ run them. These Python scripts can also optionally launch a cluster in EC2 using [boto] or in Azure using Azure CLI.
 
 Checkout [Uno] for setting up Accumulo or Fluo on a single machine.
 
-## Requirements
+## Requirements 
 
-Muchos requires the following:
+### Common
+
+Muchos requires the following common components for installation and setup:
+
+* Python 3.6.8 with a virtual environment setup
+Create a Python environment and switch to it
+```bash
+cd ~
+python3.6 -m venv env
+source env/bin/activate
+```
+* `ssh-agent` installed and running and ssh-agent forwarding. Note that this may also require the creation of SSH public-private [key pair](https://docs.microsoft.com/en-us/azure/virtual-machines/linux/mac-create-ssh-keys).
+```bash
+eval $(ssh-agent -s)
+ssh-add ~/.ssh/id_rsa
+```
+* Git (current version)
+
+### EC2
+
+Muchos requires the following for EC2 installations:
 
-* Python 3
 * [awscli] & [boto3] libraries - Install using `pip3 install awscli boto3 --upgrade --user`
-* `ssh-agent` installed and running
 * An AWS account with your SSH public key uploaded. When you configure [muchos.props], set `key.name`
  to name of your key pair in AWS.
 * `~/.aws` [configured][aws-config] on your machine. Can be created manually or using [aws configure][awscli-config].
 
+### Azure
+
+Muchos requires the following for Azure installations:
+
+* [Azure CLI](https://docs.microsoft.com/en-us/cli/azure/install-azure-cli?view=azure-cli-latest) must be installed, configured and authenticated to an Azure subscription. Please note - you should install [Azure CLI 2.0.69](https://packages.microsoft.com/yumrepos/azure-cli/azure-cli-2.0.69-1.el7.x86_64.rpm) on CentOS. Higher versions of Azure CLI are unsupported for Muchos on CentOS at this time until [this issue](https://github.com/Azure/azure-cli/issues/10128) in the Azure CLI 2.0.70 is fixed. Example command to install Azure CLI 2.0.69 on CentOS is below:
+```bash
+wget https://packages.microsoft.com/yumrepos/azure-cli/azure-cli-2.0.69-1.el7.x86_64.rpm
+sudo yum install azure-cli-2.0.69-1.el7.x86_64.rpm
+```
+* An Azure account with permissions to either use an existing or create new Resource Groups, Virtual Networks and Subnets 
+* A machine which can connect to securely deploy the cluster in Azure. If the machine is running CentOS, check if `SELinux` is enabled by running `sestatus` and if so, disable it by editing `/etc/selinux/config` file and set the SELinux mode to `disabled`. Save the file and reboot the VM.
+* Install [Ansible for Azure](https://docs.microsoft.com/en-us/azure/virtual-machines/linux/ansible-install-configure) within the Python virtual environment by using `pip install ansible[azure]`
+
 ## Quickstart
 
-The following commands will install Muchos, launch an EC2 cluster, and setup/run Accumulo:
+The following commands will install Muchos, launch a cluster, and setup/run Accumulo:
 
 ```bash
 git clone https://github.com/apache/fluo-muchos.git
+
 cd fluo-muchos/
 cp conf/muchos.props.example conf/muchos.props
 vim conf/muchos.props # Edit to configure Muchos cluster
-./bin/muchos launch -c mycluster # Launches Muchos cluster in EC2
+./bin/muchos launch -c mycluster # Launches Muchos cluster in EC2 or Azure
 ./bin/muchos setup # Set up cluster and start Accumulo
 ```
 
-The `setup` command can be run repeatedly to fix any failures and will not repeat successful operations.
+The `launch` command will create a cluster with the name specified in the command (e.g. 'mycluster'). The `setup` command can be run repeatedly to fix any failures and will not repeat successful operations.
 
 After your cluster is launched, SSH to it using the following command:
 
@@ -92,9 +124,36 @@ You can check the status of the nodes using the EC2 Dashboard or by running the
 
  ./bin/muchos status
 
+## Launching an Azure cluster
+
+Before launching a cluster, you will need to complete the requirements for Azure above, clone the Muchos repo, and
+create [muchos.props] by making a copy of existing [muchos.props.example]. If you want to give others access to your cluster, add their public keys to a file named `keys` in your `conf/` directory. During the setup of your cluster, this file will be appended on each node to the `~/.ssh/authorized_keys` file for the user set by the `cluster.username` property. You will also need to ensure you have authenticated to Azure and set the target subscription using the [Azure CLI](https://docs.microsoft.com/en-us/cli/azure/manage-azure-subscriptions-azure-cli?view=azure-cli-latest).
+
+Muchos by default uses a CentOS 7 image that is hosted in the Azure marketplace. The Azure Linux Agent is already pre-installed on the Azure Marketplace images and is typically available from the distribution's package repository. Azure requires that the publishers of the endorsed Linux distributions regularly update their images in the Azure Marketplace with the latest patches and security fixes, at a quarterly or faster cadence. Updated images in the Azure Marketplace are available automatically to customers as new versions of an image SKU.
+
+Edit the values in the sections within [muchos.props] as below
+Under the `general` section, edit following values as per your configuration
+* `cluster_type = azure`
+* `cluster_user` should be set to the name of the administrative user
+* `proxy_hostname` (optional) is the name of the machine which has access to the cluster VNET
+
+Under the `azure` section, edit following values as per your configuration 
+* `resource_group` and provide the same created in your Azure subscription for the cluster deployment
+* `vnet` and provide the same created in your Azure subscription for the cluster deployment
+* `subnet` and provide the same created in your Azure subscription for the cluster deployment
+* `numnodes` can be changed as per the cluster size 
+* `vm_sku` can be specified from the available skus in the [selected Azure region](https://docs.microsoft.com/en-us/azure/virtual-machines/linux/cli-ps-findimage)
+
+Within Azure the `nodes` section is auto populated with the hostnames and their default roles.
+
+After following the steps above, run the following command to launch an Azure VMSS cluster called `mycluster` (where 'mycluster' is the name assigned to your cluster):
+```bash
+.bin/muchos launch -c `mycluster` # Launches Muchos cluster in Azure
+```
+
 ## Set up the cluster
 
-The `./bin/muchos setup` command will set up your cluster and start Hadoop, Zookeeper, & Accumulo. It
+Once your cluster is built in EC2 or Azure, the `./bin/muchos setup` command will set up your cluster and start Hadoop, Zookeeper & Accumulo. It
 will download release tarballs of Fluo, Accumulo, Hadoop, etc. The versions of these tarballs are
 specified in [muchos.props] and can be changed if desired.
 
@@ -199,16 +258,18 @@ master, etc. It also has variables in the `[all:vars]` section that contain sett
 useful in user playbooks. It is recommended that any user-defined Ansible playbooks should be
 managed in their own git repository (see [mikewalch/muchos-custom][mc] for an example).
 
-## Terminating your EC2 cluster
+## Terminating your cluster
 
-If you launched your cluster on EC2, run the following command terminate your cluster. WARNING - All
+If you launched your cluster, run the following command to terminate your cluster. WARNING - All
 data on your cluster will be lost:
 
  ./bin/muchos terminate
 
-## Automatic shutdown of EC2 clusters
+Note: The terminate command is currently unsupported for Azure based clusters. Instead, you should delete underlying Azure VMSS resources when you need to terminate the cluster.
+
+## Automatic shutdown of clusters
 
-With the default configuration, EC2 clusters will not shutdown automatically after a delay and the default
+With the default configuration, clusters will not shutdown automatically after a delay and the default
 shutdown behavior will be stopping the node. If you would like your cluster to terminate after 8 hours,
 set the following configuration in [muchos.props]:
 
@@ -233,8 +294,10 @@ $ ./bin/muchos config -p leader.public.ip
 Muchos is powered by the following projects:
 
  * [boto] - Python library used by `muchos launch` to start a cluster in AWS EC2.
- * [Ansible] - Cluster management tool that is used by `muchos setup` to install, configure, and
+ * [ansible] - Cluster management tool that is used by `muchos setup` to install, configure, and
  start Fluo, Accumulo, Hadoop, etc on an existing EC2 or bare metal cluster.
+ * [azure-cli] - The Azure CLI is a command-line tool for managing Azure resources.
+ * [ansible-azure] - Ansible includes a suite of modules for interacting with Azure Resource Manager. 
 
 ## Muchos Testing
 
@@ -250,6 +313,8 @@ The following command runs the unit tests:
 [aws-config]: http://docs.aws.amazon.com/cli/latest/userguide/cli-config-files.html
 [awscli]: https://docs.aws.amazon.com/cli/latest/userguide/installing.html
 [awscli-config]: http://docs.aws.amazon.com/cli/latest/userguide/cli-chap-getting-started.html#cli-quick-configuration
+[azure-cli]: https://packages.microsoft.com/yumrepos/azure-cli/azure-cli-2.0.69-1.el7.x86_64.rpm
+[ansible-azure]: https://docs.ansible.com/ansible/latest/scenario_guides/guide_azure.html
 [fluo-app]: https://github.com/apache/fluo/blob/master/docs/applications.md
 [WebIndex]: https://github.com/apache/fluo-examples/tree/master/webindex
 [Phrasecount]: https://github.com/apache/fluo-examples/tree/master/phrasecount

diff --git a/ansible/accumulo.yml b/ansible/accumulo.yml
@@ -19,10 +19,10 @@
  tasks:
  - import_tasks: roles/accumulo/tasks/download.yml
  when: download_software
-- hosts: all
+- hosts: all:!{{ azure_proxy_host }}
  roles:
  - accumulo
-- hosts: accumulomaster
+- hosts: accumulomaster[0]
  tasks:
  - import_tasks: roles/accumulo/tasks/init-accumulo.yml
  handlers:

diff --git a/ansible/azure.yml b/ansible/azure.yml
@@ -0,0 +1,22 @@
+#
+# Licensed to the Apache Software Foundation (ASF) under one or more
+# contributor license agreements. See the NOTICE file distributed with
+# this work for additional information regarding copyright ownership.
+# The ASF licenses this file to You under the Apache License, Version 2.0
+# (the "License"); you may not use this file except in compliance with
+# the License. You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#
+
+
+---
+- hosts: localhost
+ roles:
+ - azure
diff --git a/ansible/common.yml b/ansible/common.yml
@@ -15,23 +15,29 @@
 # limitations under the License.
 #
 
-- hosts: nodes
- become: yes
- tasks:
- - import_tasks: roles/common/tasks/hosts.yml
 - hosts: proxy
+ become: yes
  roles:
  - proxy
  tasks:
+ - import_tasks: roles/proxy/tasks/main.yml
  - import_tasks: roles/proxy/tasks/download.yml
  when: download_software
+- hosts: nodes
+ become: yes
+ tasks:
+ - import_tasks: roles/common/tasks/hosts.yml
 - hosts: all
  become: yes
  roles:
  - common
  tasks:
  - import_tasks: roles/common/tasks/ssh.yml
  - import_tasks: roles/common/tasks/os.yml
+ - import_tasks: roles/common/tasks/azure.yml
+ when: cluster_type == 'azure'
+ - import_tasks: roles/common/tasks/azure_selinux.yml
+ when: cluster_type == 'azure'
  - import_tasks: roles/common/tasks/ec2.yml
  when: cluster_type == 'ec2'
  - import_tasks: roles/common/tasks/existing.yml

diff --git a/ansible/conf/ansible.cfg b/ansible/conf/ansible.cfg
@@ -17,3 +17,5 @@
 host_key_checking = False
 forks = 50
 gathering = smart
+callback_whitelist = profile_tasks
+timeout=30
diff --git a/ansible/docker.yml b/ansible/docker.yml
@@ -15,7 +15,7 @@
 # limitations under the License.
 #
 
-- hosts: all
+- hosts: all:!{{ azure_proxy_host }}
  become: yes
  roles:
  - docker

diff --git a/ansible/hadoop.yml b/ansible/hadoop.yml
@@ -15,12 +15,30 @@
 # limitations under the License.
 #
 
-- hosts: all
+- hosts: all:!{{ azure_proxy_host }}
  roles:
  - hadoop
+- hosts: journalnode
+ tasks:
+ - import_tasks: roles/hadoop/tasks/start-journal.yml
+- hosts: namenode[0]
+ tasks:
+ - import_tasks: roles/hadoop/tasks/format-nn.yml
+- hosts: namenode[0]
+ tasks:
+ - import_tasks: roles/hadoop/tasks/format-zk.yml
 - hosts: namenode
  tasks:
- - import_tasks: roles/hadoop/tasks/start-hdfs.yml
+ - import_tasks: roles/hadoop/tasks/start-zkfc.yml
+- hosts: namenode[0]
+ tasks:
+ - import_tasks: roles/hadoop/tasks/start-nn1.yml
+- hosts: namenode[1]
+ tasks:
+ - import_tasks: roles/hadoop/tasks/start-nn2.yml
+- hosts: workers
+ tasks:
+ - import_tasks: roles/hadoop/tasks/start-dn.yml
 - hosts: resourcemanager
  tasks:
  - import_tasks: roles/hadoop/tasks/start-yarn.yml
diff --git a/ansible/mesos.yml b/ansible/mesos.yml
@@ -15,7 +15,7 @@
 # limitations under the License.
 #
 
-- hosts: all
+- hosts: all:!{{ azure_proxy_host }}
  become: yes
  roles:
  - mesos

diff --git a/ansible/roles/accumulo/tasks/main.yml b/ansible/roles/accumulo/tasks/main.yml
@@ -59,4 +59,10 @@
  args: 
  creates: "{{ accumulo_home }}/lib/native/libaccumulo.so"
 - name: "Create accumulo log dir"
- file: path={{ worker_data_dirs[0] }}/logs/accumulo state=directory
+ file: path={{ worker_data_dirs[1] }}/logs/accumulo state=directory
+- name: "Configure max log file size to 10G for Azure log analytics integration"
+ replace:
+ path: "{{ accumulo_home }}/conf/generic_logger.xml"
+ regexp: '.*\"MaxFileSize\".*value.*'
+ replace: " <param name=\"MaxFileSize\" value=\"10000MB\"/>"
+ when: cluster_type == 'azure' and accumulo_major_version == '1'
diff --git a/ansible/roles/accumulo/templates/accumulo-env.sh b/ansible/roles/accumulo/templates/accumulo-env.sh
@@ -16,21 +16,22 @@
 # limitations under the License.
 #
 
-export ACCUMULO_LOG_DIR={{ worker_data_dirs[0] }}/logs/accumulo
+export ACCUMULO_LOG_DIR={{ worker_data_dirs[1] }}/logs/accumulo
 export ZOOKEEPER_HOME={{ zookeeper_home }}
 export JAVA_HOME={{ java_home }}
+export ACCUMULO_PID_DIR={{ worker_data_dirs[1] }}/accumulo
 
 {% if accumulo_major_version == '1' %}
 
 export HADOOP_PREFIX={{ hadoop_home }}
 export HADOOP_CONF_DIR="$HADOOP_PREFIX/etc/hadoop"
 export ACCUMULO_TSERVER_OPTS="-Xmx{{ accumulo_tserv_mem }} -Xms{{ accumulo_tserv_mem }}"
-export ACCUMULO_MASTER_OPTS="-Xmx256m -Xms256m"
-export ACCUMULO_MONITOR_OPTS="-Xmx128m -Xms64m"
-export ACCUMULO_GC_OPTS="-Xmx128m -Xms128m"
-export ACCUMULO_SHELL_OPTS="-Xmx256m -Xms64m"
+export ACCUMULO_MASTER_OPTS="-Xmx4g -Xms1g"
+export ACCUMULO_MONITOR_OPTS="-Xmx4g -Xms1g"
+export ACCUMULO_GC_OPTS="-Xmx4g -Xms1g"
+export ACCUMULO_SHELL_OPTS="-Xmx4g -Xms2g"
 export ACCUMULO_GENERAL_OPTS="-XX:+UseConcMarkSweepGC -XX:CMSInitiatingOccupancyFraction=75 -Djava.net.preferIPv4Stack=true -XX:+CMSClassUnloadingEnabled"
-export ACCUMULO_OTHER_OPTS="-Xmx256m -Xms64m"
+export ACCUMULO_OTHER_OPTS="-Xmx4g -Xms1g"
 export ACCUMULO_KILL_CMD='kill -9 %p'
 export NUM_TSERVERS=1
 export MALLOC_ARENA_MAX=${MALLOC_ARENA_MAX:-1}

diff --git a/ansible/roles/accumulo/templates/gc b/ansible/roles/accumulo/templates/gc
@@ -1 +1,3 @@
-{{ groups['accumulomaster'][0] }}
+{% for host in groups['accumulomaster'] %}
+{{ host }}
+{% endfor %}
diff --git a/ansible/roles/accumulo/templates/hadoop-metrics2-accumulo.properties b/ansible/roles/accumulo/templates/hadoop-metrics2-accumulo.properties
@@ -20,3 +20,11 @@ accumulo.sink.graphite.class=org.apache.hadoop.metrics2.sink.GraphiteSink
 accumulo.sink.graphite.server_host={{ groups['metrics'][0] }}
 accumulo.sink.graphite.server_port=2004
 accumulo.sink.graphite.metrics_prefix=accumulo
+
+{% if cluster_type == 'azure' %}
+*.sink.statsd.class=org.apache.hadoop.metrics2.sink.StatsDSink
+accumulo.sink.statsd.server.host=127.0.0.1
+accumulo.sink.statsd.server.port=8125
+accumulo.sink.statsd.skip.hostname=true
+accumulo.sink.statsd.service.name=master
+{% endif %}
diff --git a/ansible/roles/accumulo/templates/masters b/ansible/roles/accumulo/templates/masters
@@ -1 +1,3 @@
-{{ groups['accumulomaster'][0] }}
+{% for host in groups['accumulomaster'] %}
+{{ host }}
+{% endfor %}
diff --git a/ansible/roles/accumulo/templates/monitor b/ansible/roles/accumulo/templates/monitor
@@ -1 +1,3 @@
-{{ groups['accumulomaster'][0] }}
+{% for host in groups['accumulomaster'] %}
+{{ host }}
+{% endfor %}
diff --git a/ansible/roles/accumulo/templates/tracers b/ansible/roles/accumulo/templates/tracers
@@ -1 +1,3 @@
-{{ groups['accumulomaster'][0] }}
+{% for host in groups['accumulomaster'] %}
+{{ host }}
+{% endfor %}