mirror of
https://github.com/Cacti/documentation.git
synced 2026-10-30 08:37:16 +09:00
Co-authored-by: Mark Brugnoli-Vinten <netniv@hotmail.com>
129 lines
4.6 KiB
Markdown
129 lines
4.6 KiB
Markdown
# Data Settings
|
|
|
|
The Cacti Data settings includes settings for gathering statistical data
|
|
about Cacti's **Data Sources** and about settings around how Cacti
|
|
stores it's **Round Robin Archives** Time Series Databases (TSDB),
|
|
otherwise known as RRDfiles.
|
|
|
|
There are presently to sub-sections for these settings.
|
|
|
|
## Data Source Statistics
|
|
|
|
Cacti has the ability to store peak and average data for all your
|
|
**Data Sources**. You can then use that data to easily build
|
|
dashboards using third party plugins including plugins like
|
|
Superlinks from Howard Jones.
|
|
|
|
Below you will find an image of the default settings.
|
|
|
|

|
|
|
|
Those settings include:
|
|
|
|
- **Enable Data Source Statistics Collection** - This is setting
|
|
will either enable or disable the statistics collection.
|
|
|
|
- **Number of DSStats Processes** - For very large Cacti installs
|
|
the **Data Source Statistics** can take quite a long time to complete.
|
|
This setting allows Cacti to perform the statistics calculation
|
|
at orders of magnitude faster. You can allocate as many as 20
|
|
concurrent processes to perform all the updates.
|
|
|
|
- **Daily Update Frequency** - This setting determines how often
|
|
daily averages and peaks should be calculated. You can set
|
|
it to run hourly, upto every 6 hours, or after Cacti's boost
|
|
process completes.
|
|
|
|
- **Hourly Average Window** - This is a floating time window where
|
|
Cacti holds onto all samples for determining the hourly averages
|
|
for your **Data Sources**. You have to watch this cache as by
|
|
default it's stored in a memory table. You can run out of space.
|
|
|
|
- **Maintenance Time** - This is the time in the day that Cacti
|
|
will calculate Monthly and Yearly numbers and perform table
|
|
maintenance.
|
|
|
|
- **Memory Limit for Data Source Statistics Data Collector** - This
|
|
setting controls the peak memory that the statistics poller
|
|
will be allowed to use. Most systems can work fine using the
|
|
default 1GByte setting.
|
|
|
|
## RRDfile Checker
|
|
|
|

|
|
|
|
RRDfile checker scans RRDfiles and try to find problems like stale values.
|
|
|
|
Those settings include:
|
|
|
|
- **Enable RRDfile Checker** - This setting
|
|
will either enable or disable the RRD checker.
|
|
|
|
- **Number of RRDfile Checker Processes** - For very large Cacti installs
|
|
the **RRDfile Checker** can take quite a long time to complete.
|
|
This setting allows Cacti to perform the checks at orders of magnitude faster.
|
|
|
|
- **Check Frequency** - This setting determines how often
|
|
checks should be executed. You can set
|
|
it to run hourly, upto 24 hours, or after Cacti's boost
|
|
process completes.
|
|
|
|
## Data Storage Settings
|
|
|
|
These settings determine how you store your RRDfiles. This
|
|
sub-section has two layouts as showing in the images below.
|
|
|
|

|
|
|
|
The settings above include:
|
|
|
|
- **Location** - Either Local or RRDtool Proxy Server. The example
|
|
above shows the local storage option.
|
|
|
|
- **Structured RRD Paths** - In very large systems having a flat
|
|
directory structure has severe performance consequences. However,
|
|
with the Structured Paths option, Cacti will separate RRDfiles by
|
|
Device ID and use the Data Source ID as the RRDfile name.
|
|
|
|
> **NOTE**: If you are migrating from non-structured to structured
|
|
> paths setup, you must run the CLI script to move those files
|
|
> per the updated specification right after making this change.
|
|
|
|
The RRD Proxy settings below are more involved.
|
|
|
|
> **NOTE**: It should be noted that although the RRD Proxy is
|
|
> in production at large commercial sites, it continues to not
|
|
> support replication. The author claims that it's 95%
|
|
> complete, but to date, has not completed the last 5%.
|
|
|
|

|
|
|
|
The settings above include:
|
|
|
|
- **Proxy Server** - The IP or DNS name of the primary RRDfile proxy
|
|
server.
|
|
|
|
- **Proxy Port Number** - The port number to connect to in order to
|
|
communicate with the RRD Proxy.
|
|
|
|
- **RSA Fingerprint** - The fingerprint of the current public RSA
|
|
key the proxy is using. This is required to establish a trusted
|
|
connection to the RRD Proxy service.
|
|
|
|
> **WARNING**: None of the settings below are implemented.
|
|
|
|
- **Load Balancing** - Whether calls to the RRD Proxy should be
|
|
load balanced.
|
|
|
|
- **Proxy Server** - The secondary Proxy server IP or DNS name.
|
|
|
|
- **Proxy Port Numer** - The port number on the secondary RRD Proxy
|
|
to talk to.
|
|
|
|
- **RSA Fingerprint** - The fingerprint of the current public RSA
|
|
key the proxy is using. This is required to establish a trusted
|
|
connection to the RRD Proxy service.
|
|
|
|
---
|
|
<copy>Copyright (c) 2004-2022 The Cacti Group</copy>
|