Commit af2adae8 authored by Fulvio Galeazzi's avatar Fulvio Galeazzi
Browse files

Explain how the template works.

parent 720c78d8
Using the Dell MD38XX template
==============================
What is this package for
------------------------
This package contains record definition and scripts used to query
Dell MD38XX devices and gather performance statistics.
It also contains the template which should be applied to a properly configured
node.
I originally wanted to run one discovery script, create one
"virtual" host for each array and within each virtual host execute a) a
second discovery script and b) a script to fill records. Honestly, I did not
manage to achieve this so I am doing things differently.
The template works as follows:
* a discovery script finds all devices (Storage Array, RAID Controller,
Disk Group, Disk Pool). For each device a set of items is created: since
all interactions with the storage array involve running the `SMcli`
executable, which is very "heavy" and slow, the items are of type
`Zabbix trapper`, which means they are filled by a `push` from the
monitored node;
* a second discovery script is executed, to discover storage arrays whose
name matches (-m argument) some pattern;
* for each array discovered by the previous script, an item is defined
which is responsible for executing the statistics gathering script and
running `zabbix_sender`
By default, the statistics gathering script runs every 2 minutes, executing
a performance monitor session lasting 5 seconds.
Software installation and configuration
---------------------------------------
It is assumed that SM*rpm packages have already been installed.
At time of writing, and for CentOS, these are:
* SMruntime-11.25.0A06.0004-1.x86_64
* SMesm-11.25.0G06.0002-1.noarch
* SMclient-11.25.0G06.0026-1.noarch
Make your disk storage arrays known to the monitoring machine
by registering them with the command::
sudo SMcli -A <IP_address1> <IP_address2>
See what arrays have been registered with the command::
sudo SMcli -d -i
Pick one array and check whether you can successfully query
performance statistics, for example::
sudo /bin/SMcli 10.200.4.144 10.200.4.145 -e -c 'set session performanceMonitorInterval=5 performanceMonitorIterations=1 ; save storageArray performanceStats file=\"/tmp/perf.log\";'
Discovery configuration
-----------------------
Define a global regular expression to exclude devices for which you are
not interested in gathering extensive information. In our case, we exclude
the disk groups which are presented to Ceph servers, with names like `R2S1E0D3-02`
(disk group in Rack 2, Storage Array 1, Enclosure 0, Drawer 3, 2nd disk picked
from such drawer) with the expression::
Name: Dell MD38XX devices for discovery
Expression type: Result is FALSE
Expression: ^\w\d\w\d.*
In the discovery rule `DELL MD38XX discovery Devices via SMcli` we thus filter
on::
{#DEVNAME} matches @Dell MD38XX devices for discovery
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment