Page 1 of 1
Minimizing Nagios write?
Posted: Mon Feb 12, 2018 9:44 pm
by t3dus
I put Nagios Core 4.3.4 back on my Raspberry PI but and set the logs to write to memory.
My question is how do I prevent Nagios from writing to the drive constantly? I see that /usr/local/nagios/bin/nagios -d /user/local/nagios/etc/nagios.cfg is writing every 5 seconds or so.
I'm trying to maximize my SD card life. Last time my SD card only lasted 3 months cuz nagios killed it off.
Btw I'm using a Class 10 32gb Sandisk Ultra 95mb/s card so they aren't exactly super cheap.
Re: Minimizing Nagios write?
Posted: Tue Feb 13, 2018 10:33 am
by mcapra
Short list of possible ways to alleviate this depending on specifically what is causing the frequent "disk" writes, in order of "difficulty":
- Fiddle with the status_update_interval setting to write status.dat out less frequently.
- Use a ramdisk for status.dat and the checkresults file.
- Pipe logs to /dev/null, leverage the use_syslog directive to ship all Nagios logs to a syslog aggregator (this will require some rsyslog/syslog-ng knowledge). Share the resulting syslog aggregator paths as a netshare and mount to /usr/local/nagios/var/archives on the Pis. Or just don't persist logs at all and pipe everything to /dev/null (this breaks historical reporting). Or stuff them in memory as you have and not have persistence if the Pi dies.
There's a handful of places in the main
nagios.cfg file to adjust what gets written and to where:
Code: Select all
# CHECK RESULT PATH
# This is directory where Nagios stores the results of host and
# service checks that have not yet been processed.
#
# Note: Make sure that only one instance of Nagios has access
# to this directory!
check_result_path=/usr/local/nagios/var/spool/checkresults
# STATUS FILE
# This is where the current status of all monitored services and
# hosts is stored. Its contents are read and processed by the CGIs.
# The contents of the status file are deleted every time Nagios
# restarts.
status_file=/usr/local/nagios/var/status.dat
# STATUS FILE UPDATE INTERVAL
# This option determines the frequency (in seconds) that
# Nagios will periodically dump program, host, and
# service status data.
status_update_interval=10
# DEBUG VERBOSITY
# This option determines how verbose the debug log out will be.
# Values: 0 = Brief output
# 1 = More detailed
# 2 = Very detailed
debug_verbosity=1
# DEBUG FILE
# This option determines where Nagios should write debugging information.
debug_file=/usr/local/nagios/var/nagios.debug
# LOG ARCHIVE PATH
# This is the directory where archived (rotated) log files should be
# placed (assuming you've chosen to do log rotation).
log_archive_path=/usr/local/nagios/var/archives
# LOGGING OPTIONS
# If you want messages logged to the syslog facility, as well as the
# Nagios log file set this option to 1. If not, set it to 0.
use_syslog=1
I think you could get Nagios Core to keep *almost* everything in memory with clever use of a ramdisk. This can make startup a bit of a PITA, but if your only issues around stability is the cards dying I think this is a fair trade.
Re: Minimizing Nagios write?
Posted: Tue Feb 13, 2018 10:46 am
by t3dus
I have no problem manually starting services when the PI boots. The PI is on a UPS so it will rarely if ever be turned off. I'd prefer to write as much as I can to memory. I don't need historical data, this is more of a "in the now current state" monitoring that I care about.
Reason for using a PI vs one of my servers that are currently collecting dust is because I'm a cheap skate. Trying to save power and the PI is super cheap to run.
Re: Minimizing Nagios write?
Posted: Wed Feb 14, 2018 12:24 pm
by cdienger
Thanks for the excellent post
@mcapra!