Skip to content

Installation

eNMS is a Flask web application designed to run on a Unix server with Python 3.8+. The first section below describes how to get eNMS up and running quickly in demo mode to help in understanding what it is. The following sections give details for setting up a production environment.

First steps

The first step is to download the application. The user can download the latest release of eNMS directly to the browser by going to the Release section of eNMS github repository.

The other option is to clone the master branch of the git repository from github:

# download the code from github:
git clone https://github.com/eNMS-automation/eNMS.git
cd eNMS

Once the application is installed, the user must go to the eNMS folder and install eNMS python dependencies:

# install the requirements:
pip install -r build/requirements/requirements.txt

Once the requirements have been installed, the user can run the application with the Flask built-in development server.

# set the FLASK_APP environment variable
export FLASK_APP=app.py

# start the application with Flask
flask run --host=0.0.0.0

For use in a production environment, it is advised to create a more robust environment for the application. The following sections describe the recommended requirements:

Database

By default, eNMS will use SQLite database. The user can configure a different database from SQLAlchemy's list of supported databases

For example, for a MySQL database, change the DATABASE_URL environment variable:

export DATABASE_URL="mysql://root:password@localhost/enms"

Flask Secret key

Provide a secret key used by Flask to sign sessions:

# set the SECRET_KEY environment variable
export SECRET_KEY=value-of-user-secret-key

Server Address

The address is needed when eNMS needs to provide a link back to the application, which is the case with webssh and mail notifications. When left empty, eNMS will try to guess the URL. This might not work consistently depending on the user's environment (nginx configuration, proxy, ...).

export SERVER_ADDR="http://192.168.56.102"

WSGI server

A WSGI HTTP server such as gunicorn is required to run eNMS in production, instead of the Flask development server.

A recommended configuration file for gunicorn is in the main folder: gunicorn.py; it is recommended to run the application with the following command:

# start the application with gunicorn
gunicorn --config gunicorn.py app:app

Dramatiq

Dramatiq, a distributed task queue, can be used for executing automations:

  1. In setup/settings.json set "use_task_queue": true
  2. Set the REDIS_ADDR environment variable and run dramatiq eNMS from the project root.

Hashicorp Vault

All credentials should be stored in a Hashicorp Vault: the settings variable use_vault : true under the vault section of the setup/settings.json file tells eNMS that a vault has been setup. Follow the manufacturer instructions and options for how to setup a Hashicorp Vault

Tell eNMS how to connect to the Vault with environment variables:

  • VAULT_ADDRESS
  • VAULT_TOKEN

eNMS can also unseal the Vault automatically at start time. This mechanism is disabled by default. To activate it, one must set unseal_vault : true in setup/settings.json and set the UNSEAL_VAULT_KEY environment variables :

export UNSEAL_VAULT_KEY1=key1
export UNSEAL_VAULT_KEY2=key2
etc

Plugin Installation

Any initial eNMS Plugins - like the sample eNMS CLI Plugin - can also be installed here. See Plugins for more details.

Environment variables

Environment variables for all sensitive data (passwords, tokens, keys) are exported from Unix and include:

  • SECRET_KEY=secret_key
  • MAIL_PASSWORD=mail_password
  • TACACS_PASSWORD=tacacs_password
  • SLACK_TOKEN=slack_token

Example Systemd Unit and Socket files and Nginx config with proxy-pass

enms.gunicorn.socket

[Unit]
Description=gunicorn socket for enms

[Socket]
ListenStream=/run/gunicorn/enms.gunicorn.socket

[Install]
WantedBy=sockets.target

enms.gunicorn.service

[Unit]
Description=Gunicorn instance to serve enms
Requires=enms.gunicorn.socket
Requires=vault.service
After=network.target
After=mysqld.service
After=enms.gunicorn_scheduler.service
After=vault.service

[Service]
PermissionsStartOnly=true
PIDFile=/run/gunicorn/enms.gunicorn.pid
User=centos
Group=centos
WorkingDirectory=/home/centos/enms
ExecStartPre=/bin/mkdir -p /run/gunicorn/
ExecStartPre=/bin/chown -R centos:centos /run/gunicorn/
# Add the virtualenv to the PATH
Environment="PATH=/opt/python3-virtualenv/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:."
ExecStartPre=/bin/echo Setting application PATH to $PATH
Environment="http_proxy=http://proxy.company.com:80/"
Environment="https_proxy=http://proxy.company.com:80/"
Environment="NO_PROXY=localhost,127.0.0.1,169.254.169.254,.company.com,W.X.Y.Z/8,A.B.C.D/16"
Environment="HOSTNAME=HOSTNAME_CHANGEME"
Environment="SECRET_KEY=SECRET_KEY_CHANGEME"
Environment="VAULT_ADDR=http://127.0.0.1:8200"
Environment="VAULT_LOG_LEVEL=info"
Environment="VAULT_TOKEN=VAULT_TOKEN_CHANGEME"
Environment="UNSEAL_VAULT_KEY1=VAULT_KEY1_CHANGEME"
Environment="UNSEAL_VAULT_KEY2=VAULT_KEY2_CHANGEME"
Environment="UNSEAL_VAULT_KEY3=VAULT_KEY3_CHANGEME"
Environment="SCHEDULER_ADDR=http://127.0.0.1:8000"
# To indicate which server a service was run from in Automation/Run table
Environment="SERVER_NAME=CHANGE_ME"
Environment="SERVER_ADDR=CHANGE_ME"
# Use Fernet Key additional encryption for stored passwords - not compatible with loading
# eNMS example migration files (comment out for examples)
Environment="FERNET_KEY=SOME_FERNET_VALID_KEY"
Environment="ENMS_ADDR=https://127.0.0.1"
Environment="ENMS_PASSWORD=ENMS_PASSWORD_CHANGEME"
Environment="REDIS_ADDR=127.0.0.1"
Environment="LDAP_ADDR=ldap://MYCOMPANY_CHANGEME"
Environment="LDAP_USERDN=MYCOMPANY_CHANGEME"
Environment="LDAP_BASEDN=DC=my,DC=ad,DC=company,DC=com"
Environment="TACACS_PASSWORD=CHANGEME"
Environment="MAIL_PASSWORD=CHANGEME"
Environment="SLACK_TOKEN=CHANGEME"
Environment="FLASK_APP=app.py"
Environment="FLASK_DEBUG=1"
# PATH to TextFSM repo for Netmiko: https://pynet.twb-tech.com/blog/automation/netmiko-textfsm.html
Environment="NET_TEXTFSM=/home/centos/ntc_textfsm/"
Environment="GUNICORN_ACCESS_LOG=logs/access.log"
Environment="GUNICORN_LOG_LEVEL=debug"
Environment="DATABASE_URL=mariadb+mysqldb://root:PASSWORD@localhost/enms?charset=utf8mb4"
ExecStart=/opt/python3-virtualenv/bin/bin/gunicorn --pid /run/gunicorn/enms.gunicorn.pid --worker-tmp-dir /tmpfs-gunicorn --bind unix:/run/gunicorn/enms.gunicorn.socket --chdir /home/centos/enms --config /home/centos/enms/gunicorn.py app:app
ExecReload=/bin/kill -s HUP $MAINPID
ExecStop=/bin/kill -s TERM $MAINPID
TimeoutStopSec=60
LimitNOFILE=100000

[Install]
WantedBy=multi-user.target

nginx.conf

user nginx;
worker_processes auto;
error_log /var/log/nginx/error.log;
pid /run/nginx.pid;

# Load dynamic modules. See /usr/share/nginx/README.dynamic.
include /usr/share/nginx/modules/*.conf;

events {
    worker_connections 1024;
}

http {
    log_format  main  '$remote_addr - $remote_user [$time_local] "$request" '
                  '$status $body_bytes_sent "$http_referer" '
                  '"$http_user_agent" "$http_x_forwarded_for"';

    access_log  /var/log/nginx/access.log  main;
    sendfile            on;
    tcp_nopush          on;
    tcp_nodelay         on;
    keepalive_timeout   120;
    types_hash_max_size 2048;
    fastcgi_buffers 8 16k;
    fastcgi_buffer_size 32k;
    fastcgi_connect_timeout 900;
    fastcgi_send_timeout 900;
    fastcgi_read_timeout 900;

    include             /etc/nginx/mime.types;
    default_type        application/octet-stream;

    # Load modular configuration files from the /etc/nginx/conf.d directory.
    # See http://nginx.org/en/docs/ngx_core_module.html#include
    # for more information.
    include /etc/nginx/conf.d/*.conf;

    server {
        # port to listen on. Can also be set to an IP:PORT
        listen 443 http2 ssl default_server;
        listen [::]:443 http2 ssl default_server;
        server_name enms-hub;
        server_name_in_redirect on;
        server_tokens off;
        ssl_certificate /etc/ssl/certs/nginx.pem;
        ssl_certificate_key /etc/ssl/private/nginx.key;
        ssl_dhparam /etc/ssl/certs/dhparam.pem;
        client_max_body_size 0;
        client_body_buffer_size 10K;
        client_header_buffer_size 1k;

        ssl_protocols TLSv1.1 TLSv1.2;
        ssl_prefer_server_ciphers on;
        ssl_ciphers "EECDH+AESGCM:EDH+AESGCM:AES256+EECDH:AES256+EDH";
        ssl_ecdh_curve secp384r1;
        ssl_buffer_size 16k;
        ssl_session_cache shared:SSL:10m;
        ssl_session_timeout 4h; # 4 hours to reuse sessions
        ssl_session_tickets off;
        ssl_stapling off;
        ssl_stapling_verify off;

        # Disable preloading HSTS for now.  One can use the commented out header line that includes
        # the "preload" directive if one understands the implications.
        #add_header Strict-Transport-Security "max-age=63072000; includeSubdomains; preload";
        add_header Strict-Transport-Security "max-age=63072000; includeSubdomains";
        add_header X-Frame-Options DENY;
        add_header X-Content-Type-Options nosniff;

        root  /usr/share/nginx/html;

        # Load configuration files for the default server block.
        #include /etc/nginx/default.d/*.conf;

        # Enable compression for larger files/results
        gzip on;
        gzip_types      text/plain application/xml application/json text/json text/xml;
        gzip_proxied    no-cache no-store private expired auth;
        gzip_min_length 100000;
        gunzip on;

        location / {
            proxy_set_header Host $http_host;
            proxy_set_header X-Real-IP $remote_addr;
            proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
            proxy_set_header X-Forwarded-Proto $scheme;
            proxy_pass http://unix:/run/gunicorn/enms.gunicorn.socket;
            proxy_http_version 1.1;
            proxy_set_header Upgrade $http_upgrade;
            proxy_set_header Connection "upgrade";
            proxy_connect_timeout 300s;
            # Should match the gunicorn timeout:
            proxy_read_timeout 7200s;
        }

        # Use regex for terminal urls:
        location ~ ^/terminal(.*)$ {
            proxy_set_header X-Real-IP $remote_addr;
            proxy_set_header X-Forwarded-For $remote_addr;
            proxy_set_header Host $host;        
            rewrite ^/terminal(\d*)/(.*)$ /$2 break;
            proxy_pass http://127.0.0.1:$1;
            proxy_http_version 1.1;
            proxy_set_header Upgrade $http_upgrade;
            proxy_set_header Connection "upgrade";
        }

        error_page 404 /404.html;
            location = /40x.html {
        }
        error_page 500 502 503 504 /50x.html;
            location = /50x.html {
        }

        # Setup enms Documentation link
        location /docs/ {
            proxy_set_header Host $http_host;
            proxy_set_header X-Real-IP $remote_addr;
            proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
            proxy_set_header X-Forwarded-Proto $scheme;
        }
    }
    # Added to expose access to the separate/remote scheduler.
    server {
        listen 8000;
        listen [::]:8000;
        server_name enms-scheduler;
        root  /usr/share/nginx/html;
        location / {
             proxy_set_header Host $http_host;
             proxy_set_header X-Real-IP $remote_addr;
             proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
             proxy_set_header X-Forwarded-Proto $scheme;
             proxy_pass http://unix:/run/gunicorn/enms.gunicorn_scheduler.socket;
             proxy_connect_timeout 300s; 
             # Should match the gunicorn timeout:
             proxy_read_timeout 7200s;
        }
    }
}

Note

Importing the files/migrations/examples migration files requires that the FERNET_KEY environment variable be unset (not defined) and that the hash_user_passwords parameter in setup/settings.json be set to True because of how the sample passwords are stored in the examples. The files/migrations/default migration files that most production deployments start with do not contain passwords, so loading them does not impose the same requirements.

Note

Deploying eNMS into an environment that uses proxy url redirection requires that the http_proxy=, https_proxy=, and NO_PROXY= environment variables be set if that is needed to reach certain endpoints, such as the network_data repository for device configuration data.

Settings Files

The application's setup files are located in setup/, and the following section describes their contents:

automation.json

The setup/automation.json file allows for the customization of some automation library options, such as Napalm Getters, the default contents of the parameterized run form for workflows, the default arguments used to open a Scrapli connection, and the models that one can access or create from the workflow builder global variables.

database.json

The setup/database.json file contains database and schema configuration parameters. Make sure to include a new database engine section here if switching to another SQLAlchemy supported database variant.

Key parameters to be aware of:

  • pool_size (default: 1000) Number of connections kept persistently in SQL Alchemy pool.
  • max_overflow (default: 10) Maximum overflow size of the connection pool.
  • tiny_string_length (default: 64) Length of a tiny string in the database.
  • small_string_length (default: 255) Length of a small string in the database.
  • small_string_length (default: 32768) Length of a large string in the database.

logging.json

Logging settings exist in the file setup/logging.json. This file is directly passed into the Python Logging library, so it uses the Python3 logger file configuration syntax for the user's version of Python3. Using this file, the administrator can configure additional loggers and logger destinations as needed for workflows.

By default, the two loggers are configured:

  • The default logger has handlers for sending logs to the stdout console as well as a rotating log file logs/enms.log.
  • A security logger captures logs for: User A ran Service/Workflow B on Devices [C,D,E...] to log file logs/security.log.

And these can be reconfigured here to forward through syslog to remote collection if desired. This can also be used as a means to send workflow data to remote time series databases for dashboarding and longer term storage of automation results data.

Additionally, the external loggers section allows for changing the log levels for the various libraries used by eNMS.

With multiple gunicorn workers, please consider:

  • Using Python WatchedFileHandler instead of the RotatingFileHandler.
  • Configuring the LINUX logrotate utility to perform the desired log rotation.

Example logging.json that allows for logging to a time series database in the workflow's post-processing section or via python snippet service: 

{
  "version": 1,
  "disable_existing_loggers": false,
  "formatters": {
    "standard": {
      "format": "%(asctime)s %(levelname)-8s %(message)s"
    },
    "rsyslogdFormatter": {
      "format": "%(filename)s: %(funcName)s: %(message)s"
    },
    "timeseriesSyslogFormatter": {
      "format": "logtype=\"workflow\" %(message)s"
    },
  },
  "handlers": {
    "console": {
      "level": "DEBUG",
      "formatter": "standard",
      "class": "logging.StreamHandler",
      "stream": "ext://sys.stdout"
    },
    "rotation": {
      "level": "DEBUG",
      "formatter": "standard",
      "filename": "logs/enms.log",
      "class": "logging.handlers.WatchedFileHandler"
    },
    "security": {
      "level": "DEBUG",
      "formatter": "rsyslogdFormatter",
      "class": "logging.handlers.SysLogHandler",
      "address": "/dev/log",
      "facility": "local3"
    },
    "timeseries": {
      "level": "INFO",
      "formatter": "timeseriesSyslogFormatter",
      "class": "logging.handlers.SysLogHandler",
      "address": ["X.X.X.X", PortYYYY]
    },
  },
  "loggers": {
    "": {
      "handlers": ["console", "rotation"],
      "level": "DEBUG"
    },
    "security": {
      "handlers": ["security"],
      "level": "DEBUG",
      "change_log": true,
      "service_log": true
    },
    "timeseries": {
      "handlers": ["timeseries"],
      "level": "INFO"
    },
  },
  "external_loggers": {
    "engineio": "error",
    "requests": "info",
    "socketio": "error",
    "urllib3": "info",
    "werkzeug": "error"
  }
}

properties.json

The setup/properties.json file includes:

  1. Allowing for additional custom properties to be defined in eNMS for devices and links. In this way, eNMS device inventory can be extended to include additional columns/fields.
  2. Allowing for additional custom parameters to be added to services and workflows.
  3. Controlling which parameters and widgets can be seen from the Dashboard.
  4. Controlling which column/field properties are visible in the tables for device and link inventory, configuration, pools, as well as the service, results, and task browsers.
  5. Pull-down choices for Category, Model, Operating System, and Vendor.

For examples, please see Custom Properties for more information

rbac.json

The setup/rbac.json file allows configuration of which user roles have access to each of the controls in the UI, as well as which user roles have access to each of the REST API endpoints.

settings.json

The setup/settings.json file includes the following public variables which are also modifiable from the administration panel. Changing settings from the administration panel will cause settings.json to be rewritten if Write changes back to settings.json file is selected.

app section

  • config_mode (default: "debug") Must be set to "debug" or "production".
  • documentation_url (default: "https://enms.readthedocs.io/en/latest/") Can be changed if one wants to host one's own version of the documentation locally. Points to the online documentation by default.
  • git_repository (default: "") Git is used as a version control system for device configurations: this variable is the address of the remote git repository where eNMS will push all device configurations.
  • plugin_path: (default: "eNMS/plugins") location of eNMS plugin extensions and customizations.
  • session_timeout_minutes: (default: 90).
  • startup_migration (default: "examples") Name of the migration to load when eNMS starts for the first time.
    • By default, when eNMS loads for the first time, it will create a network topology and a number of services and workflows as examples of what the user can do.
    • One can set the migration to "default" instead, in which case eNMS will only load what is required for the application to function properly.

authentication section

Lists the methods available for users to login (in order of which they are shown in the login screen pulldown list). Default authentication is specified here. Because authentication can be custom for many environments, eNMS/custom.py allows the user to customize how the authentication needs to occur. It can be modified to fit a company's ldap active directory system, etc.

automation section

  • max_process limit on multiprocessing (default: 15).

cluster section

Section used for detecting other running instances of eNMS. - active (default: false). - id (default: true). - scan_subnet (default: "192.168.105.0/24"). - scan_protocol (default: "http"). - scan_timeout (default: 0.05).

docs section

This section is used to configure which pages in the documentation to open for some of the embedded information (help) links. This should not need to be changed.

mail section

  • reply_to (default: "no_replies@company.com").
  • server (default: "smtp.googlemail.com").
  • port (default: 587).
  • use_tls (default: true).
  • username (default: "eNMS-user").
  • sender (default: "eNMS@company.com").

mattermost section

  • url (default: "https://mattermost.company.com/hooks/i1phfh6fxjfwpy586bwqq5sk8w").
  • channel (default: "").
  • verify_certificate (default: true).

paths section

  • files (default:"") Path to eNMS managed files needed by services and workflows. For example, files to upload to devices.
  • custom_code (default: "") Path to custom libraries that can be utilized within services and workflows.
  • custom_services (default: "") Path to a folder that contains Custom Services. These services are added to the list of existing services in the Automation Panel when building services and workflows.
  • playbooks (default: "") Path to where Ansible playbooks are stored so that they are selectable in the Ansible Playbook service.

redis section

This section allows configuration of the Redis queue.

  • charset (default:"utf-8").
  • db (default:0).
  • decode_responses (default:true).
  • port (default:6379).
  • socket_timeout (default:0.1).

requests section

Allows for tuning of the Python Requests library internal structures for connection pooling. Tuning these might be necessary depending on the load on eNMS.

  • Pool

  • pool_maxsize (default: 10).

  • pool_connections (default: 100).

  • pool_block (default: false).

  • Retries

  • total (default: 2).

  • read (default: 2).
  • connect (default: 2).
  • backoff_factor (default: 0.5).

security section

  • hash_user_passwords (default: true) All user passwords are automatically hashed by default.
  • forbidden_python_libraries (default: ["eNMS","os","subprocess","sys"]) There are a number of places in the UI where the user is allowed to run custom python scripts. The user can configure which python libraries cannot be imported for security reasons.

slack section

  • channel (default: "").

ssh section

  • command (default: python3 -m flask run -h 0.0.0.0): command used to start the SSH server.

  • credentials - which credential types to enable for the WebSSH connection feature.

    • custom (default: true).
    • device (default: true).
    • user (default: true).

  • port_redirection (default: false).

  • bypass_key_prompt (default: true).
  • port (default: -1).
  • start_port (default: 9000).
  • end_port (default: 9100).

tables section

Configure the refresh rates of a table in the UI in milliseconds. By default, the "Results" page (run table) will be refreshed every 5 seconds, and the "Services" and "Tasks" pages every 3 seconds.

  • run (default: 5000).
  • service (default: 3000).
  • task (default: 3000).

vault section

For eNMS to use a Vault to store all sensitive data (user and network credentials), one must set the active variable to true, provide an address and export:

  • active (default: false).
  • unseal (default: false) Automatically unseal the Vault.

The keys must be exported as environment variables:

  • VAULT_ADDRESS
  • VAULT_TOKEN
  • UNSEAL_VAULT_KEY1
  • UNSEAL_VAULT_KEY2
  • UNSEAL_VAULT_KEY3
  • UNSEAL_VAULT_KEY4
  • UNSEAL_VAULT_KEY5

themes.json

The setup/themes.json file exposes some configurable parameters for the default and dark appearance themes of the application.

visualization.json

The setup/visualization.json file controls the visualization map portion of the application. The user can specify whether the visualization defaults to '2D' mode (recommended for large inventories) or '3D' mode. Also, the default latitude and longitude of the map can be set here.

Key parameters to note:

  • default ("2D" or "3D").
  • longitude (default: -96.0).
  • latitude (default: 33.0).
  • zoom_level (default: 5).
  • tile_layer (default: "osm").
  • marker (default: "Circle").

Key parameters for the network builder:

  • display_nodes_as_images (true): displaying nodes as images can be slower if there are too many nodes.
  • max_allowed_nodes (500): threshold above which eNMS will not try to display a network (too much data).

Scheduler

The scheduler, used for running tasks at a later time, is a web application that is distinct from eNMS. It can be installed on the same server as eNMS, or a remote server.

Before running the scheduler, one must configure the following environment variables so it knows where eNMS is located and what credentials to authenticate with:

  • ENMS_ADDR: URL of the remote server (example: "http://192.168.56.102").
  • ENMS_USER: eNMS login.
  • ENMS_PASSWORD: eNMS password.

The scheduler is an asynchronous application that must be deployed with gunicorn :

cd scheduler
gunicorn scheduler:scheduler --host 0.0.0.0

Example Systemd Unit and Socket files

Above nginx.conf sample has a section for eNMS Scheduler

enms.gunicorn_scheduler.socket

[Unit]
Description=Gunicorn socket for eNMS scheduler

[Socket]
ListenStream=/run/gunicorn/enms.gunicorn_scheduler.socket

[Install]
WantedBy=sockets.target

enms.gunicorn_scheduler.service

[Unit]
Description=Start eNMS Scheduler service using Gunicorn
Requires=enms.gunicorn_scheduler.socket
After=network.target
After=mysqld.service

[Service]
PermissionsStartOnly=true
PIDFile=/run/gunicorn/enms.gunicorn_scheduler.pid
User=centos
Group=centos
WorkingDirectory=/home/centos/enms/eNMS
ExecStartPre=/bin/mkdir -p /run/gunicorn/
ExecStartPre=/bin/chown -R centos:centos /run/gunicorn/
Environment="ENMS_ADDR=https://127.0.0.1"
Environment="ENMS_PASSWORD=CHANGE_ME"
Environment="ENMS_USER=admin"
Environment="VERIFY_CERTIFICATE=0"
Environment="REDIS_ADDR=127.0.0.1"
Environment="SCHEDULER_ADDR=http://127.0.0.1:8000"
Environment="GUNICORN_ACCESS_LOG=/home/centos/enms/logs/access_scheduler.log"
Environment="GUNICORN_LOG_LEVEL=info"
Environment="DATABASE_URL=mysql://root:PASSWORD@localhost/enms?charset=utf8"
ExecStart=/opt/python3-virtualenv/bin/gunicorn --pid /run/gunicorn/enms.gunicorn_scheduler.pid --worker-tmp-dir /tmpfs-gunicorn --bind unix:/run/gunicorn/enms.gunicorn_scheduler.socket --chdir /home/centos/enms/scheduler --config /home/centos/enms/scheduler/gunicorn.py scheduler:scheduler
ExecReload=/bin/kill -s HUP $MAINPID
ExecStop=/bin/kill -s TERM $MAINPID
TimeoutStopSec=60
LimitNOFILE=100000

[Install]
WantedBy=multi-user.target

scheduler.json

The setup/scheduler.json file controls the behavior of the eNMS Scheduler application.

Network Data Merge Driver

eNMS features easy access to up-to-date device configurations and network data. A service can be configured to periodically collect the latest configurations direct from devices in a network, after which a backend process can leverage Git to synchronize data between all instances of the application. These files can be found in the network_data/ repository of each VM, organized by device name.

Services can also be configured to update only their instance's local repository. This intended functionality has been known to cause merge conflicts during automatic synchronizations with the shared remote repository.

To prevent this, a git merge driver can be used to programmatically resolve merge conflicts. Such drivers typically utilize a combination of git attributes, configuration instructions, and custom scripts to function. Below are all of the components of a proposed driver for the Network Data repositories:

network_data/.gitattributes

...
* merge=newest
...

This line in the attributes file instructs git to associate the merge method named "newest" for every file (through the wildcard character) in the repository.

network_data/.git/config

...
[merge "newest"]
    name = Merge by newest commit
    driver = git-merge-newest %O %A %B %L %P master FETCH_HEAD
[branch "master"]
    mergeoptions = --no-edit
...

When added to the existing .git/config, these lines define a new merge strategy and provide the necessary command line arguments, as well as several placeholder values handled by git. More detailed descriptions of these values (along with additional resources on custom merge drivers in general) are available in Git's documentation.

network_data/git-merge-newest

#!/bin/sh
if git merge-file -p -q "$2" "$1" "$3" > /dev/null;
    then git merge-file "$2" "$1" "$3";
else
    MINE=$(git log --format="%ct" --no-merges "$6" -1);
    THEIRS=$(git log --format="%ct" --no-merges "$7" -1);
    if [ $MINE -gt $THEIRS ]; then
        git merge-file -q --ours "$2" "$1" "$3" >/dev/null
    else
        git merge-file -q --theirs "$2" "$1" "$3">/dev/null
    fi
fi

The shell script defines the logic and commands to run when the merge driver is called. In this case it compares the timestamps from git log commands on the local master branch and the contents of FETCH_HEAD, accepting the most recent result.