Skip to main content

Documentation Portal

Linux Deployment

Linux Deployment

TrendMiner can be installed on CentOS 7 and Red Hat Enterprise Linux 7. Other distributions based on Red Hat Enterprise Linux 7 might work but are not officially supported. During the installation and any future updates, access to the official repositories or a mirror of them is required.

TrendMiner has no extra requirements when installing on a customer-provided CentOS 7 or Red Hat Enterprise Linux 7 system outside of the software that is available in the official operating system repositories.

Prepare the Installation

Allocate Storage TrendMiner stores application data and logs in /mnt/data. Ensure that this directory has sufficient free disk space.

Important

The TrendMiner appliance uses the following IP addresses/ranges:

These IP addresses should not be used by other applications/servers running on the network on which TrendMiner is deployed. If your company is already using one or more of these IP addresses, please contact TrendMiner support before installing.

Firewall

TrendMiner controls the firewall of the system. Additional ports that need to be open for monitoring or other agents need to be configured as documented in Opening Additional Firewall Ports.

Install TrendMiner

  1. Copy the installation archive to the destination server:

    $ scp <installation file> <user>@<server>:<path>

  2. Extract the installation archive:

    # tar -xf <installation file>

  3. Run the installation script and send the output to a log file. Choose any location to save this log file:

    # ./install-trendminer.sh -s server_name | tee trendminer-installation.log

    Important

    The ‘server_name’ has to be the FQDN excluding the protocol of the server. An example: ./install-trendminer.sh -s my_tm_url.com

    Note that the FQDN is case sensitive.

  4. Choose the correct installation method.

    What do you want to do:

    [1]Install Single node
[2]Install Cluster
[3]Deploy application
->

  5. Type 1 to start the single node installation. Wait for the installation to finish. Continue by configuring TrendMiner in ConfigHub.

Post-Installation Configuration

Opening Additional Firewall Ports TrendMiner controls the packet filtering rules of the Linux kernel using iptables. By default, only icmp, ssh (tcp/22) and http(s) (tcp/80, tcp/443) are allowed.

Additional ports can be opened by creating a configuration file, owned by root:

/etc/ansible/facts.d/iptables.fact

This file contains JSON in the following format:

{
    "allow_icmp": true,
    "open_tcp": [
        5666,
        10050
    ]
}

In this example, ports 5666 and 10050 are opened on the TrendMiner instance to allow respectively NRPE and Zabbix connectivity on the system.

Post-Installation Configuration

Validate the JSON syntax by running:

# python -m json.tool < /etc/ansible/facts.d/iptables.fact

This will echo the configuration on success. In case of errors, it will print the location:

Expecting , delimiter: line 5 column 9 (char 65)

To apply the new firewall configuration without downtime, execute:

# cd /mnt/data/ansible-playbooks
# ansible-playbook agent/configure-tm-agent.yml \
    -i /var/www/dsquare-appliance/hosts

The updated configuration will be applied on subsequent reboots and preserved across future TrendMiner updates.

Monitor TrendMiner

TrendMiner collects metrics into a Prometheus instance that is running on every TrendMiner system. Prometheus metrics exporters included in TrendMiner include:

  • Node Exporter exports OS level metrics

  • Postgres Exporter exports database metrics

  • Each TrendMiner service exports JVM metrics

  • Several TrendMiner services export application specific metrics

All metrics are available for external systems through Prometheus federation. Alerts based on these metrics can then be created in an external monitoring stack.

By default the metrics are not exposed. To access them, first allow port 9100 as described in the Opening Additional Firewall Ports section. Then scrape the /federate endpoint of the Prometheus instance.

When using a monitoring stack based on Prometheus and AlertManager, Prometheus could be configured as:

scrape_configs:
  - job_name: federated-trendminer
    honor_labels: true
    honor_timestamps: true
    scrape_interval: 15s
    scrape_timeout: 10s
    metrics_path: /federate
    static_configs:
      - targets:
         - <my-trendminer>:9100

AlertManager can then handle alerts:

groups:
  - name: TrendMiner rules
    rules:
      - alert: instance_down_2m
        expr: up{job="federated-trendminer"} == 0
        for: 2m
        labels:
          severity: Critical
        annotations:
          description: TrendMiner DOWN for 2 minutes
          summary: TrendMiner DOWN
In this section: