Table of Contents

• Installation
  • Get the Code
  • Install Vagrant
    • Install VirtualBox and Vagrant
      • MacOS
      • Linux (Ubuntu)
• Using Monasca Vagrant
  • Starting mini-mon
  • Basic Monasca usage
  • Smoke test
  • Updating
  • Running behind a Web Proxy
  • Running with Vertica
• Advanced Usage
  • Access information
    • Internal Endpoints
  • Improving Provisioning Speed
• Monasca Debugging
  • Ansible Development
    • Running Ansible directly
    • Editing Ansible Configuration
• Developing Monasca
• Alternate Vagrant Configurations
• Troubleshooting

Installs a mini monitoring environment based on Vagrant. Intended for development of the monitoring infrastructure.

Installation

Get the Code

git clone https://github.com/stackforge/monasca-vagrant

Install Vagrant

Install VirtualBox and Vagrant

Note: Vagrant version 1.5.0 or higher is required.

MacOS

The following steps assume you have Homebrew installed. Otherwise, install VirtualBox and Vagrant and Ansible as suggested on their websites.

brew tap phinze/cask
brew install brew-cask
brew cask install virtualbox 
brew cask install vagrant
brew install ansible  # Version 1.8+ is required
ansible-galaxy install -r requirements.yml -p ./roles

Linux (Ubuntu)

sudo apt-get install virtualbox
#Download and install latest vagrant from http://www.vagrantup.com/downloads.html
sudo pip install ansible  # Version 1.8+ is required
ansible-galaxy install -r requirements.yml -p ./roles

Using Monasca Vagrant

Starting mini-mon

• After installing to start just run vagrant up. The first run will download required vagrant boxes.
• When done you can run vagrant halt to stop the boxes and later run vagrant up to turn them back on. To destroy and rebuild run vagrant destroy -f. It is typically fastest to use halt/up than to rebuild your vm.
• Run vagrant help for more info on standard vagrant commands.

Basic Monasca usage

The full Monasca stack is running on the mini-mon vm and many devstack services on the devstack vm. A monasca-agent is installed on both and metrics are actively being collected.

• You can access the horizon UI by navigating to http://192.168.10.5 and logging in as mini-mon/password. This is the UI used for devstack and it contains the Monasca plugin found at the Monitoring tab as well as Grafana used for graphing metrics.
• Run vagrant ssh <host> to log in, where <host> is either mini-mon or devstack
• The monasca cli is installed within both vms and the necessary environment variables loaded into the shell. This is a good way to explore the metrics in the system. For example to list all metrics, run monasca metric-list

Smoke test

At the end of the install a smoke test is run that exercises every major piece of Monasca. If this fails the end of the provision will report it. It is possible to rerun this at any point using Ansible ansible-playbook ./smoke.yml or from within the vm by running smoke.py and smoke2.py in /opt/monasca/hpcloud-mon-monasca-ci\*/tests/smoke.

Updating

When someone updates the config, this process should allow you update the VMs, though not every step is needed at all times.

• git pull
• ansible-galaxy install -r requirements.yml -p ./roles -f
• vagrant box update Only needed rarely
• vagrant provision, if the vms where halted run vagrant up first.
  • It is also possible to Ansible directly to update just parts of the system. See Ansible Development for more info.

Running behind a Web Proxy

If you are behind a proxy you can install the vagrant-proxyconf plugin to have Vagrant honor standard proxy-related environment variables and set the VM to use them also. It is important that 192.168.10.4, 192.168.10.5, 127.0.0.1 and localhost be in your no_proxy environment variable.

vagrant plugin install vagrant-proxyconf

Running with Vertica

You can configure Vagrant to run Vertica as the database in place of influxdb.

To accomplish this you have to download the community edition (Debian) and the jdbc driver from Vertica.

Place the jdbc driver and debian in the home directory of vagrant with the names of:

vertica_jdbc.jar vertica.deb

Set the environment variable USE_VERTICA to true and then run vagrant up.

export USE_VERTICA=true
vagrant up

Advanced Usage

Access information

• Your host OS home dir is synced to /vagrant_home on the VM.
• The root dir of the monasca-vagrant repo on your host OS is synced to /vagrant on the VM.
• mini-mon is at 192.168.10.4 and devstack is at 192.168.10.5

Internal Endpoints

• Influxdb web ui is available at http://192.168.10.4:8083 with root/root as user/password
• The Monasca-api is available at http://192.168.10.4:8080
  • The keystone credentials used are mini-mon/password in the mini-mon project. The keystone services on 192.168.10.5 on standard ports.

Improving Provisioning Speed

The slowest part of the provisioning process is the downloading of packages. The Vagrant plugin vagrant-cachier available at https://github.com/fgrehm/vagrant-cachier should help by caching repeated dependencies. To use with Vagrant simply install the plugin.

sudo vagrant plugin install vagrant-cachier

Monasca Debugging

See this page for details on the Monasca Architecture.

The components of the system which are part of the Monasca code base have there configuration in /etc/monasca and their logs in /var/log/monasca. For nearly all of these you can set the logging to higher debug level and restart. The components of the system which are dependencies for Monasca (zookeeper, kafka, storm, influxdb, mysql) are either in the standard Ubuntu location or in /opt.

Some other helpful commands:

• Zookeeper shell at - /usr/share/zookeeper/bin/zkCli.sh
• Kafka debug commands are at /opt/kafka/bin in particular the kafka-console-consumer.sh is helpful.
• Running monasca-collector info will give an report on the current state of agent checks.
• The storm admin webui exists at http://192.168.10.4:8088
• The mysql admin is root/password so you can access the db with the command mysql -uroot -ppassword mon

Ansible Development

Running Ansible directly

At any point you can rerun vagrant provision to rerun the Ansible provisioning. Often it is easier to run ansible directly and specify tags, ie ansible-playbook mini-mon.yml --tags api,persister. Also a very simple playbook is available for running the smoke test, ansible-playbook ./smoke.yml

For these to work smoothly add these vagrant specific settings to your local ansible configuration (~/.ansible.cfg or a personal ansible.cfg in this dir):

[defaults]
hostfile = .ansible_hosts

# In some configurations this won't work, use only if your config permits.
[ssh_connection]
pipelining = True  # Speeds up connections but only if requiretty is not enabled for sudo

Next run vagrant ssh-config >> ~/.ssh/config, that will set the correct users/host_keys for the vagrant vms.

When running Ansible directly make sure that you pass in what the database_type is, ie ansible-playbook mini-mon.yml -e 'database_type=influxdb'.

Editing Ansible Configuration

Since there are only two VMs in this setup the Ansible configuration has no host or group variables, rather all variables are in the playbook. There is one playbook for each machine, mini-mon.yml and devstack.yml. The playbooks contain all variables, some tasks and the roles used in building the VMs.

To edit the Ansible roles I suggest downloading the full git source of the role and putting it in your ansible path. This allows you to run your changes directly from the git copy you are working on. See the Ansible docs for more details on the exact configuration needed.

Developing Monasca

In this repo there are a couple of helper scripts to aid in downloading all of the Monasca git repositories.

• ./monasca-repos.sh <parent_dir> will clone all code repos to the parent dir
• ./monasca-ansible-repos.sh <parent_dir> will clone all of the team Ansible repos to the parent dir

Alternate Vagrant Configurations

To run any of these alternate configs, simply run the Vagrant commands from within the subdir.

• ds-build subdir - This is used for building a new devstack server image. It does not typically need to be run.

Previously in the split directory an alternative setup was available with each service split into different vms and using Vertica rather than influxdb. This was removed simply because it was not being actively maintained as changes occurred. It is still possible to split up the services and to use Vertica, these are done in test environments and production deployments, however is beyond the scope of this development environment. Additionally, other alternative setups including running mini-mon in HP Public Cloud and scripts for putting it on baremetal are also no longer supported.

Troubleshooting

1. Service Unavailable (HTTP 503) - If you receive an error like below when provisioning, then ensure your proxy config is setup as shown in Running behind a Web Proxy TASK: [monasca-keystone | Keystone Service - Execute the script] ************** (stacktrace elided...) keystoneclient.openstack.common.apiclient.exceptions.ServiceUnavailable: Service Unavailable (HTTP 503)