-
Language:
English
-
Language:
English
Metrics Store Installation Guide
Installing Metrics Store for Red Hat Virtualization
Abstract
Preface
Metrics Store collects logs and metrics from Red Hat Virtualization. The data is transferred from Red Hat Virtualization to Red Hat OpenShift where it is stored and aggregated in Elasticsearch and saved in indexes. The data can then be analyzed and visualized in Kibana.
- Elasticsearch is a distributed, RESTful search and analytics engine that lets you perform and combine many types of searches.
- Kibana is an open source analytics and visualization platform designed to work with Elasticsearch. You can easily perform advanced data analysis and visualize your data in a variety of charts and tables.
Chapter 1. Workflow and architecture
To install Metrics Store, complete the following major tasks:
Metrics Store architecture
The Metrics Store architecture is based on the Red Hat OpenShift EFK logging stack, running on Red Hat OpenShift Container Platform 3.11.
Metrics Store uses the following services:
- collectd (hosts) collects metrics from hosts, virtual machines, and databases.
- rsyslog (hosts) collects metrics, adds log data, enriches the data with metadata, and sends the enriched data to Elasticsearch.
- Elasticsearch (Metrics Store virtual machine) stores and indexes the data.
- Kibana (Metrics Store virtual machine) analyzes and presents the data as dashboards and charts.
Figure 1.1. Metrics Store architecture
Chapter 2. Installing Metrics Store
Prerequisites
Computing resources:
- 4 CPU cores
- 30 GB RAM
- 500 GB SSD disk
For the Metrics Store Installer virtual machine:
- 4 CPU cores
- 8 GB RAM
NoteThe computing resource requirements are for an all-in-one installation, with a single Metrics Store virtual machine. The all-in-one installation can collect data from up to 50 hosts, each running 20 virtual machines.
- Operating system: Red Hat Enterprise Linux 7.7 or later
- Software: Red Hat Virtualization 4.3.5 or later
- Network configuration: see Configuring networking for Metrics Store virtual machines
2.1. Creating the Metrics Store virtual machines
To create the Metrics Store virtual machines, perform the following tasks:
- Configure the Metrics Store installation.
Create the following Metrics Store virtual machines:
- The Metrics Store Installer virtual machine - a temporary virtual machine for deploying Red Hat OpenShift and services on the Metrics Store virtual machines.
- One or more Metrics Store virtual machines.
- Verify the Metrics Store virtual machines.
2.1.1. Configuring the Metrics Store installation
Procedure
- Log in to the Manager machine using SSH.
Update the packages:
# yum update
Copy
metrics-store-config.yml.example
to createmetrics-store-config.yml
:# cp /etc/ovirt-engine-metrics/metrics-store-config.yml.example /etc/ovirt-engine-metrics/config.yml.d/metrics-store-config.yml
-
Edit the parameters in
metrics-store-config.yml
to match your installation environment, and save the file. The parameters are documented in the file. To set the logical network that is used for the
metrics-store-installer
and Metrics Store virtual machines, add the following lines tometrics-store-config.yml
:# ovirt_template_nics - the following are the default values for setting the logical network used by the metrics_store_installer and the Metrics Store virtual machines ovirt_template_nics: - name: nic1 profile_name: ovirtmgmt interface: virtio
On the Manager machine, copy
/etc/ovirt-engine-metrics/secure_vars.yaml.example
to/etc/ovirt-engine-metrics/secure_vars.yaml
:# cp /etc/ovirt-engine-metrics/secure_vars.yaml.example /etc/ovirt-engine-metrics/secure_vars.yaml
-
Edit the parameters in
/etc/ovirt-engine-metrics/secure_vars.yaml
to match the details of your specific environment. Encrypt the
secure_vars.yaml
file:# ansible-vault encrypt /etc/ovirt-engine-metrics/secure_vars.yaml
2.1.2. Creating Metrics Store virtual machines
Procedure
Go to the
ovirt-engine-metrics
directory:# cd /usr/share/ovirt-engine-metrics
Run the
ovirt-metrics-store-installation
playbook to create the virtual machines:# ANSIBLE_JINJA2_EXTENSIONS="jinja2.ext.do" ./configure_ovirt_machines_for_metrics.sh --playbook=ovirt-metrics-store-installation.yml --ask-vault-pass
NoteTo enable verbose mode for debugging, add
-vvv
to the end of the command, or add '-v' to enable light verbose mode, or add-vvvv
to enable connection debugging. For more extensive debugging options, enable debugging through the Ansible playbook as described in Enable debugging via Ansible playbook
2.1.3. Verifying the creation of the virtual machines
Procedure
- Log in to the Administration Portal.
-
Click Compute → Virtual Machines to verify that the
metrics-store-installer
virtual machine and the Metrics Store virtual machines are running.
2.1.4. Changing the default LDAP authentication identity provider (optional)
In the standard Metrics Store installation, the allow_all
identity provider is configured by default. You can change this default during installation by configuring the openshift_master_identity_providers
parameter in the inventory file integ.ini
.
You can also configure the session options in the OAuth configuration in the integ.ini
inventory file.
Procedure
-
Locate the
integ.ini
in the root directory of themetrics-store-installer
virtual machine. - Follow the instructions for updating the identity provider configuration in Configuring identity providers with Ansible.
2.2. Configuring networking for Metrics Store virtual machines
2.2.1. Configuring DNS resolution for Metrics Store virtual machines
Procedure
-
In the
metrics-store-config.yml
DNS zone parameter,public_hosted_zone
should be defined as a wildcard DNS record (*.example.com
). That wildcard DNS should resolve to the IP address of yourmaster0
virtual machine. - Add the hostnames of the Metrics Store virtual machines to your DNS server.
2.2.2. Setting a static MAC address for a Metrics Store virtual machine (optional)
Procedure
- Log in to the Administration Portal.
- Click Compute → Virtual Machines and select a Metrics Store virtual machine.
- In the Network Interfaces tab, select a NIC and click Edit.
- Select Custom MAC Address, enter the MAC address, and click OK.
- Reboot the virtual machine.
2.2.3. Configuring firewall ports
The following table describes the firewall settings needed for communication between the ports used by Metrics Store.
Table 2.1. Configure the firewall to allow connections to specific ports
ID | Port(s) | Protocol | Sources | Destinations | Purpose |
---|---|---|---|---|---|
MS1 | 9200 | TCP |
RHV Red Hat Virtualization Hosts | Metrics Store VM | Transfer data to ElasticSearch. |
MS2 | 5601 | TCP | Kibana user | Metrics Store VM | Give users access to the Kibana web interface. |
Whether a connection is encrypted or not depends on how you deployed the software.
2.3. Deploying Metrics Store services on Red Hat OpenShift
Deploy and verify Red Hat OpenShift, Elasticsearch, Curator (for managing Elasticsearch indices and snapshots), and Kibana on the Metrics Store virtual machines.
Procedure
-
Log in to the
metrics-store-installer
virtual machine. Run the
install_okd
playbook to deploy Red Hat OpenShift and Metrics Store services to the Metrics Store virtual machines:# ANSIBLE_CONFIG="/usr/share/ansible/openshift-ansible/ansible.cfg" \ ANSIBLE_ROLES_PATH="/usr/share/ansible/roles/:/usr/share/ansible/openshift-ansible/roles" \ ansible-playbook -i integ.ini install_okd.yaml -e @vars.yaml -e @secure_vars.yaml --ask-vault-pass
NoteTo enable verbose mode for debugging, add
-vvv
to the end of the command, or add '-v' to enable light verbose mode, or add-vvvv
to enable connection debugging.Verify the deployment by logging in to each Metrics Store virtual machine:
Log in to the
openshift-logging
project:# oc project openshift-logging
Check that the Elasticsearch, Curator, and Kibana pods are running:
# oc get pods
If Elasticsearch is not running, see Troubleshooting related to ElasticSearch in the OpenShift Container Platform 3.11 documentation.
Check the Kibana host name and record it so you can access the Kibana console in Chapter 4, Verifying the Metrics Store installation:
# oc get routes
Cleanup
- Log in to the Administration Portal.
- Click Compute → Virtual Machines and delete the metrics-store-installer virtual machine.
Chapter 3. Deploying collectd and rsyslog
Deploy collectd
and rsyslog
on the hosts to collect logs and metrics.
You do not need to repeat this procedure for new hosts. The Manager configures the hosts automatically.
Procedure
- Log in to the Manager machine using SSH.
Copy
/etc/ovirt-engine-metrics/config.yml.example
to create/etc/ovirt-engine-metrics/config.yml.d/config.yml
:# cp /etc/ovirt-engine-metrics/config.yml.example /etc/ovirt-engine-metrics/config.yml.d/config.yml
Edit the
ovirt_env_name
andelasticsearch_host
parameters inconfig.yml
and save the file. These parameters are mandatory and are documented in the file.NoteIf you add a Manager or an Elasticsearch installation, copy the Manager’s public key to your Metrics Store virtual machine using the following commands:
# mytemp=$(mktemp -d) # cp /etc/pki/ovirt-engine/keys/engine_id_rsa $mytemp # ssh-keygen -y -f $mytemp/engine_id_rsa > $mytemp/engine_id_rsa.pub # ssh-copy-id -i $mytemp/engine_id_rsa.pub root@{elasticsearch_host} # rm -rf $mytemp
Deploy
collectd
andrsyslog
on the hosts:# /usr/share/ovirt-engine-metrics/setup/ansible/configure_ovirt_machines_for_metrics.sh
Chapter 4. Verifying the Metrics Store installation
Verify the Metrics Store installation using the Kibana console. You can view the collected logs and create data visualizations.
Procedure
Log in to the Kibana console using the URL (
https://kibana.example.com
) that you recorded in Section 2.3, “Deploying Metrics Store services on Red Hat OpenShift”. Use the defaultadmin
user, and the password you defined during the metrics store installation.Optionally, you can access the Red Hat OpenShift Container Platform portal at
https://example.com:8443
(using the sameadmin
user credentials).Select the Discover tab, and check that you can view the project.ovirt-logs-ovirt_env_name-uuid index.
See the Discover section in the Kibana User Guide for information about working with logs.
Select the Visualize tab, where you can create data visualizations for the project.ovirt-metrics-ovirt_env_name-uuid and the project.ovirt-logs-ovirt_env_name-uuid indexes.
The Metrics Store User Guide describes the available parameters. See the Visualize section of the Kibana User Guide for information about visualizing logs and metrics.
Appendix A. Installing Metrics Store with Satellite
You can use Satellite to install Metrics Store on a disconnected environment.
Prerequisites
- The Satellite server is configured. For more information, see Disconnected installation using Satellite Docker registry
If you encounter a missing image or a reference to an online image (depending on which applications you are using), consider updating the references in the deployment or build configuration of the application, or re-tag Docker images as a temporary measure (just to rule out that the image is not reachable).
The following
OpenShift
component images are synchronized through Docker on your Satellite server:openshift3/oauth-proxy openshift3/ose-console openshift3/ose-control-plane openshift3/ose-deployer openshift3/ose-docker-registry openshift3/ose-haproxy-router openshift3/ose-logging-auth-proxy openshift3/ose-logging-curator5 openshift3/ose-logging-elasticsearch5 openshift3/ose-logging-fluentd openshift3/ose-logging-kibana5 openshift3/ose-node openshift3/ose-pod openshift3/ose-web-console openshift3/registry-console rhel7/etcd
Two hosts are created on the Satellite server - one for the Metrics Store Installer virtual machine, and one for the OpenShift virtual machine, as follows:
- Create hosts on Satellite - see Creating a Host.
-
Assign static IP addresses and MAC addresses for the virtual machines. The host for the OpenShift virtual machine should be of the format
master-<suffix>0
to match the OpenShift virtual machine hostname.
The
qcow
image is available on the Manager machine.- Go to RHEL product software.
-
In the Product Software tab, download the
Red Hat Enterprise Linux KVM Guest Image
to the Manager machine.
Running the Ansible role
On the Manager machine, copy
/etc/ovirt-engine-metrics/metrics-store-config-satellite.yml.example
tometrics-store-config.yml
.# cp /etc/ovirt-engine-metrics/metrics-store-config-satellite.yml.example /etc/ovirt-engine-metrics/config.yml.d/metrics-store-config.yml
Update the values of
/etc/ovirt-engine-metrics/metrics-store-config.yml
to match the details of your specific environment.# vi /etc/ovirt-engine-metrics/config.yml.d/metrics-store-config.yml
On the Manager machine, copy
/etc/ovirt-engine-metrics/secure_vars_satellite.yaml.example
to/etc/ovirt-engine-metrics/secure_vars.yaml
.# cp /etc/ovirt-engine-metrics/secure_vars_satellite.yaml.example /etc/ovirt-engine-metrics/secure_vars.yaml
Update the values of /etc/ovirt-engine-metrics/secure_vars.yaml to match the details of your specific environment.
# vi /etc/ovirt-engine-metrics/secure_vars.yaml
Encrypt the
secure_vars.yaml
file.# ansible-vault encrypt /etc/ovirt-engine-metrics/secure_vars.yaml
Go to the ovirt-engine-metrics repo.
# cd /usr/share/ovirt-engine-metrics
Run the metrics store installation playbook that creates the metrics store installer virtual machine.
# ANSIBLE_JINJA2_EXTENSIONS="jinja2.ext.do" ./configure_ovirt_machines_for_metrics.sh \ --playbook=ovirt-metrics-store-installation.yml --ask-vault-pass -vvv
- Log in to the Administration Portal and review the Metrics Store installer virtual machine creation.
Log in to the Metrics Store installer virtual machine.
# ssh root@<metrics-store-installer ip or fqdn>
Run the Ansible playbook that deploys OpenShift on the virtual machines that were created.
# ANSIBLE_CONFIG="/usr/share/ansible/openshift-ansible/ansible.cfg" \ ANSIBLE_ROLES_PATH="/usr/share/ansible/roles/:/usr/share/ansible/openshift-ansible/roles" \ ansible-playbook -i integ.ini install_okd.yaml -e @vars.yaml -e @secure_vars.yaml --ask-vault-pass -vvv
Appendix B. Sharing reports with non-administrators
Users without administrator privileges can view collected logs and metrics as read-only users. The following example creates a user named user name with view (read-only) permissions.
Procedure
- Log in to the Metrics Store virtual machine.
Create a new user:
# oc create user username # oc create identity allow_all: username # oc create useridentitymapping allow_all: username username
Log in to the openshift-logging project:
# oc project openshift-logging
Assign a view role to the user:
# oc adm policy add-role-to-user view user name
Create a password for the user:
# oc login --username=user name --password=password
Appendix C. Removing Metrics Store
To remove the Metrics Store installation:
Stop and disable
collectd
andrsyslog
:# cd /usr/share/ovirt-engine-metrics/ # ./configure_ovirt_machines_for_metrics.sh --playbook=cleanup-ovirt-metrics.yml -vvv
- Delete the Metrics Store virtual machine.