A personal snippets to run distributed LibreNMS poller on low-end server resources to monitor hundreds of devices and sensors.
A normal LibreNMS install contains all parts of LibreNMS: - Poller/Discovery workers - RRD (Time series data store) - Database - Webserver (Web UI/API)
Devices monitored with LibreNMS can be grouped together into a
poller_group to pin these devices to a single or a group of designated pollers.
Distributed Polling allows the workers to be spread across additional servers for horizontal scaling. All pollers need to write to the same set of RRD files, shared NFS storage can be used, but I prefer using RRDcached.
It is a required that all pollers can access the central memcached to communicate with each other.
Actually, the distributed poller setup is very dynamic and can be configured to fit your need. Below are my current setup on what you need to consider both from the software layer but also connectivity.
|Node||OS||RAM - CPU||Services||Devices|
|nms82||Ubuntu 18.4||4GB - 4 Core||Web/API, MariaDB, RRDcached, Redis, Memcached, Syslog||< 9|
|nms83||Ubuntu 18.4||4GB - 4 Core||Poller, Oxidized Config Backup||62|
|nms84||Ubuntu 18.4||4GB - 4 Core||Poller, Discovery Module||19|
|bsd-007-c7||CentOS 7||2GB - 4 Core||Poller||128|
All machines above is run under Proxmox Linux Container across different location.
For now, this resource (4 core Xeon and 4GB RAM each node) seem to be overkill to monitor about 200 devices, 1800 ports and 1000 sensors including processor, storages, applications, and disk I/O. The master host (nms82) only use about 1.5% of CPU resource and 600 - 1.200 MiB of RAM.
But you may be need to switch from 7,000 RPM HDD to 15,000 RPM SAS drive or SSD for better performance since disk write is pretty high.
All this setup need a working LibreNMS install, plus:
- Python 3
These can be obtained from OS package manager, or from PyPI with command below (under librenms root directory):
1pip3 install -r requirements.txt
Master Host (Web UI, API, MariaDB, Redis, RRDcached, Memcached) (nms82)
This node become one of most important part which contain additional required server running on this node: Web UI / API, Memcached, RRDcached and Redis server.
Web / API Layer
Based on topology above, the web / API layer will respond to Nginx reverse proxy. This is typically Apache but I prefer using Nginx, here some Nginx backend configuration example for LibreNMS.
The pollers, web and API layers should all be able to access the database server directly. Make sure that MariaDB listen to server IP address that can be accessed from other nodes instead of unix socket.
Additionally, you can use stable version of official MariaDB repo instead of distribution repo.
RRD Storage (via RRDCached)
It is advisable to run RRDCached within this setup so that you don’t need to share the rrd folder via a remote file share such as NFS. The web service can then generate rrd graphs via RRDCached.
For this example, I’m running RRDCached to allow all pollers and web/api servers to read/write to the rrd files with the rrd directory.
1apt install rrdcached
1DAEMON=/usr/bin/rrdcached 2WRITE_THREADS=4 3BASE_PATH=/opt/librenms/rrd/ 4JOURNAL_PATH=/var/lib/rrdcached/journal/ 5PIDFILE=/var/run/rrdcached.pid 6SOCKFILE=/var/run/rrdcached.sock 7SOCKGROUP=librenms 8DAEMON_GROUP=librenms 9DAEMON_USER=librenms 10NETWORK_OPTIONS="-L" 11BASE_OPTIONS="-B -F -R"
Change ownership of journal path to
1chown librenms:librenms /var/lib/rrdcached/journal/
Memcache is required for the distributed pollers to be able to register to a central location and record what devices are polled. Memcache can run from any of the servers so long as it is accessible by all pollers.
Example memcached config on ubuntu
1# memcached default config file 2# 2003 - Jay Bonci <[email protected]> 3# This configuration file is read by the start-memcached script provided as 4# part of the Debian GNU/Linux distribution. 5 6# Run memcached as a daemon. This command is implied, and is not needed for the 7# daemon to run. See the README.Debian that comes with this package for more 8# information. 9-d 10 11# Log memcached's output to /var/log/memcached 12logfile /var/log/memcached.log 13 14# Be verbose 15# -v 16 17# Be even more verbose (print client commands as well) 18# -vv 19 20# Start with a cap of 64 megs of memory. It's reasonable, and the daemon default 21# Note that the daemon will grow to this size, but does not start out holding this much 22# memory 23-m 64 24 25# Default connection port is 11211 26-p 11211 27 28# Run the daemon as root. The start-memcached will default to running as root if no 29# -u command is present in this config file 30-u memcache 31 32# Specify which IP address to listen on. The default is to listen on all IP addresses 33# This parameter is one of the only security measures that memcached has, so make sure 34# it's listening on a firewalled interface. 35#-l 127.0.0.1 36-l 0.0.0.0 37 38# Limit the number of simultaneous incoming connections. The daemon default is 1024 39# -c 1024 40 41# Lock down all paged memory. Consult with the README and homepage before you do this 42# -k 43 44# Return error when memory is exhausted (rather than removing items) 45# -M 46 47# Maximize core file limit 48# -r 49 50# Use a pidfile 51-P /var/run/memcached/memcached.pid
Note: You also need
Don’t forget to configure server firewall so, only selecred server and poller can access the memcached server.
Redis instance needed to coordinate the nodes. It’s recommended that you do not share the Redis database with any other system. By default, Redis supports up to 16 databases (numbered 0-15). On this topic, I use database number 4 for LibreNMS and use
chris-lea PPA for Redis 3+ on Ubuntu:
1add-apt-repository ppa:chris-lea/redis-server 2apt install redis-server
Edit Ubuntu Redis configuration example: (
1bind 0.0.0.0 2requirepass YOURSCRETREDISPASSWORD
It’s strongly recommended that you deploy a resilient cluster of redis systems, and use
Poller Hosts (nms83, nms84, bsd-006-c7)
All poller host need working LibreNMS install. All you need to do after that is reconfigure all poller host config connection (Redis, Database, RRDCached, Memcached) to connect to Master Host (nms82).
Connection Configuration on All Nodes
Once all required dependency is installed and running on master hosts, reconfigure it in the .env and config.php file on each node.
Connection settings are required in
.env file is generated after
composer install and
NODE_ID are set. Remember that the
APP_KEY value must be the same on all pollers and leave
NODE_ID as it.
1#APP_KEY= #Required, generated by composer install 2 3DB_HOST=[YOUR_DATABASE_HOST] 4DB_DATABASE=[YOUR_DATABASE_NAME] 5DB_USERNAME=[YOUR_DATABASE_USER] 6DB_PASSWORD=[YOUR_DATABASE_PASS] 7 8#NODE_ID= #Required, generated by composer install 9LIBRENMS_USER=librenms 10 11# Distributed Polling 12REDIS_HOST=[YOUR_REDIS_HOST_IP] 13REDIS_PORT=6379 14REDIS_PASSWORD=[YOURSCRETREDISPASSWORD] 15REDIS_DB=4 16CACHE_DRIVER=redis
config.php under LibreNMS directory :
1<?php 2 3$config['distributed_poller'] = true; 4$config['distributed_poller_name'] = php_uname('n'); 5$config['distributed_poller_group'] = '0'; # Set this to your prefered group for each node 6 7$config['distributed_poller_memcached_host'] = "[IP_ADDR_OF_MEMCACHED_SERVER]"; 8$config['distributed_poller_memcached_port'] = 11211; 9 10 11$config['service_poller_workers'] = 24; # Processes spawned for polling 12$config['service_services_workers'] = 8; # Processes spawned for service polling 13$config['service_discovery_workers'] = 16; # Processes spawned for discovery 14 15 16//Optional Settings 17#$config['service_poller_frequency'] = 300; # Seconds between polling attempts 18#$config['service_services_frequency'] = 300; # Seconds between service polling attempts 19#$config['service_discovery_frequency'] = 21600; # Seconds between discovery runs 20#$config['service_billing_frequency'] = 300; # Seconds between billing calculations 21#$config['service_billing_calculate_frequency'] = 60; # Billing interval 22#$config['service_poller_down_retry'] = 60; # Seconds between failed polling attempts 23#$config['service_loglevel'] = 'WARNING'; # Must be one of 'DEBUG', 'INFO', 'WARNING', 'ERROR', 'CRITICAL' 24#$config['service_update_frequency'] = 86400; # Seconds between LibreNMS update checks 25 26$config['service_watchdog_enabled'] = true;
A systemd unit file is provided - the sysv and upstart init scripts could also be used with a little modification.
A systemd unit file can be found in
misc/librenms.service. To install run
cp /opt/librenms/misc/librenms.service /etc/systemd/system/librenms.service && systemctl enable --now librenms.service.