Table of Contents |
---|
...
Forewarned is forearmed the poverb goes, a quick google tells me "prior knowledge of possible dangers or problems gives one a tactical advantage". The reason we want to baseline and threshold our data is so that we can receive alerts forewarning us of issues in our environment, so that we can act to resolve smaller issues before they become bigger. Being proactive increases our Mean Time Between Failure.
...
What is really key here is that the baseline tool will detect downward changes as well as upward changes, so if your traffic was reducing outside the baseline you would be alerted.
Establishing a Dynamic Baseline
Current Value
Firstly I want to calculate my current value, I could use the last value collected, but depending on the stability of the metric this might cause false positives, as NMIS has always supported, using a larger threshold period when calculating the current value can result in more relevant results.
...
If this was a weekly pattern the multi-day baseline would be a better option, but if this happens more randomly, using the same-day would generate an initial event on the increase, then the event would clear as the ~8Mbps became normal, and then when the value dropped again another alert would be generated.
Delta Baseline
The delta baseline is only concerned with the amount of change in the baseline, for example
Working with the Dynamic Baseline and Thresholding Tool
...
Configuration Option | Description | Example |
---|---|---|
baseline | Which type of baseline are we using, "dynamic" or "delta", the default is dynamic, if undefined, dynamic will be used. | delta |
active | Is baselining this metric active or not, values are true or false | true |
metric | Which NMIS data point or variable, equates to an RRD DS | RouteNumber |
type | Which NMIS model section or metric | RouteNumber |
section | What is the section name in the node info, just run it, otherwise the section must exist. | |
nodeModel | This is a regex which defines which NMIS models should be matched | CiscoRouter |
event | The name of the event to use, will default to Proactive Baseline type metric if none provided. | Proactive Route Number Change |
indexed | Is this variable indexed or not | false |
threshold_exceeds | Ignored if undef otherwise the value must ALSO exceed this threshold to raise an event | undef |
threshold_period | How many minutes should the value to be baselined be averaged, e.g. -5 minutes is the last poll, -15 minutes would be the average of the last 15 minutes, -1 hour would be the last 60 minutes. | -5 minutes |
multiplier | How many standard deviations to vary the baseline by. | 1 |
weeks | The number of weeks to look back | 0 |
hours | The number of hours to include in the baseline metrics | 8 |
levels | The levels section is used by the delta baseline method to define when an amount of change will trigger an event and what level that event will be. |
Same-Day Dynamic Baseline Configuration Example
Here is what the configuration file would look like, this example is a Same-Day Baseline:
Code Block |
---|
'RouteNumber' => { 'active' => 'true', 'metric' => 'RouteNumber', 'type' => 'RouteNumber', 'nodeModel' => 'CiscoRouter', 'event' => 'Proactive Route Number Change', 'indexed' => 'false', 'threshold_exceeds' => undef, 'threshold_period' => "-5 minutes", 'multiplier' => 1, 'weeks' => 0, 'hours' => 8, }, |
Multi-Day Dynamic Baseline Configuration Example
Another configuration option using the BGP Prefixes being exchanged with BGP peers, is from systemHealth modelling and this is a multi-day baseline:
Code Block |
---|
'cbgpAcceptedPrefix' => {
'active' => 'true',
'metric' => 'cbgpAcceptedPrefix',
'type' => 'bgpPrefix',
'section' => 'bgpPrefix',
'nodeModel' => 'CircuitMonitor|CiscoRouter',
'event' => 'Proactive BGP Peer Prefix Change',
'indexed' => 'true',
'multiplier' => 1,
'weeks' => 4,
'hours' => 1,
}, |
Delta Baseline Configuration Example
Currently delta baselines do not support multi-day, but the hours value can be very large if required.
Code Block |
---|
'hrSystemProcesses' => {
'baseline' => 'delta',
'active' => 'true',
'metric' => 'hrSystemProcesses',
'type' => 'Host_Health',
'nodeModel' => 'net-snmp',
'indexed' => 'false',
'hours' => 4,
'levels' => {
'Warning' => 10,
'Minor' => 20,
'Major' => 30,
'Critical' => 40,
'Fatal' => 50
}
}, |
Running the Baseline Tool
...
Additional options will be added, running the tool with no arguments will tell you the currently supported options.
Command Line options for Node and Group
To have the tool only run for a subset of devices you can use node_regex and group_regex options. These are useful for only running the tool for a single node while testing new baseline configurations or in the case of the group_regex, you may only require the baseline tool to run for a subset of your devices.
Running for a couple of nodes using regular expressions.
Code Block |
---|
/usr/local/omk/bin/baseline.exe act=run node_regex="router1|server2" |
Running for a couple of groups using regular expressions.
Code Block |
---|
/usr/local/omk/bin/baseline.exe act=run group_regex="HQ|Data Center|West Coast" |
Automatic Processing using Cron
The baseline tool should have created a cron.d configuration /etc/cron.d/baseline, which will contain the following.
Code Block |
---|
# # this cron schedule runs the baseline system every 5 minutes. # # # if you DON'T want any NMIS cron mails to go to root, # uncomment and adjust the next line #MAILTO=prefered@domain.com # # m h dom month dow user command # # run the baseline every 5 minutes starting at 4 minutes offset from the hour. 4-59/5 * * * * root /usr/local/omk/bin/baseline.exe act=run > /usr/local/omk/log/baseline.log 2>&1 |
Using Group Regex and Cron for Parallel Processing.
The group regex option can be used to provide parallel processing if the baseline tool is taking longer than 5 minutes to run. A simple example would be using the baseline tool for all core and distribution devices in one processing run and a second one for all access devices.
Code Block |
---|
# run the baseline every 5 minutes starting at 3 and 4 minutes offset from the hour.
3-58/5 * * * * root /usr/local/omk/bin/baseline.exe act=run group_regex="Core|Dist" > /usr/local/omk/log/baseline1.log 2>&1
4-59/5 * * * * root /usr/local/omk/bin/baseline.exe act=run group_regex="Access" > /usr/local/omk/log/baseline2.log 2>&1 |