[Feature Request] Additional reporting metrics

[RobertT]

Would it be possible to add some additional reporting metrics/graphs?
Specifically I would like temperature. Preferably for Motherboard, CPU and disks.

I found this plugin: https://github.com/lunetics/collectd-plugins/tree/master/ipmitool
I am able to get to work with my hardware on 9.2.1.8 to collect data but the data isnt graphed.
It seems to me that it would be easy to extend that design to include data from smartctl output as well.

What is the best way to get something like that included in future releases?
Or is there some other way I can collect+graph this data that would not be wiped out after an upgrade?

 

[dlavigne]

Please create a feature request at bugs.freenas.org and post the issue number here.

 

[cyberjock]

There's probably a 1/2 dozen on this already, so please make sure a ticket doesn't exist currently before creating a new one. I'd almost bet dollars to donuts one is currently open for this.

The "best" way to get this feature (or any feature) added is to do it yourself. The developers definitely appreciate code that works already. ;)

Starting with 9.2.1.7 your graph data should never be wiped out if you are saving the RRD data to the pool via the checkbox in the WebGUI.

 

[RobertT]

I am sure you are probably right that there is a ticket but I apparently suck at searching for it..

If I am going to try to add the feature myself I really would want to talk with someone who is a frequent contributor in order to get advice. It would be better to get some advice on implementation rather than to put something together and not have it accepted upstream.

 

[cyberjock]

As someone who isn't a developer or anything, but has made a few patches to FreeNAS, they appreciate everything. As long as it works that's about all they care about. They will look over the code to make sure there's no glaring bugs, secret back doors, etc. but they always accept it unless there's a specific reason they will never allow a feature. For instance, if you were to add some feature that is a major security risk and they have expressly stated this, then they might not accept it. But I haven't personally seem them reject anything.

 

[Nindustries]

I'd love this!

 

[depasseg]

It looks like SMART drive temp monitoring got pushed to the next major OS release (from 9.3-Beta). https://bugs.freenas.org/issues/147

I couldn't find a CPU or System Temp bug. Is there any chance you are trying to add this capability to your own system? :)

 

[Fox]

FWIW, there is a script for drive temps.. It could be graphed...

https://forums.freenas.org/index.php?threads/python-script-to-monitor-drive-temps.22794/

 

[RobertT]

So the current graphs in freenas are generated by collectd. collectd has various plugins available to it. One plugin (that already exists on freenas) is called "exec". That plugin can be configured to gather data collected from a shell script. There is a script here: https://github.com/lunetics/collectd-plugins/tree/master/ipmitool that will poll the ipmi values on the motherboard for the CPU info and allow collectd to create the rrd for the data. In my initial messing around with this I was able to collect the rrd by running my own seperate collectd daemon with my own config file. When I started looking at the required changes to the freenas gui to show the graphs I sort of lost interest. The problem to me there was that all the graphs used custom settings for their rrd output and I didnt feel like mucking around in rrd config directives just to look at pretty cpu / disk temperature graphs.
Perhaps someone more interested could take the concept farther and actually get other graphs integrated into freenas. Or better yet someone could recode the freenas reporting to allow a "plugin" style customization.

I dont know if anyone out there has actaully used the original sun unified that had ZFS under the hood and saw how the reporting was done on it. That was the most awesome reporting I have ever seen on a storage array. For those of you that are curious here are a couple of links..
(must see video even if you dont really care about the analytics)
http://dtrace.org/resources/bmc/cec_analytics.pdf

If you are curious about the custom collectd config I used here you go:
Hostname "localhost"
FQDNLookup true
BaseDir "/var/db/collectd"
PIDFile "/var/run/collectd-custom.pid"
PluginDir "/usr/local/lib/collectd"

LoadPlugin syslog
LoadPlugin rrdtool
LoadPlugin exec

<Plugin "syslog">
LogLevel err
</Plugin>

<Plugin "rrdtool">
DataDir "/var/db/collectd/rrd"
CacheTimeout 120
CacheFlush 900
</Plugin>

<Plugin exec>
Exec "myipmi" "/data/mine/collectd-ipmitool.sh"
</Plugin>

 

[Apollo]

Interesting video. I guess they could use their fancy server system to detect earthquakes too.
I haven't looked at the RRD function, but I know/I think Pfsense does use RRDTool to graph data as well without loss of granularity over time.
As stated by Cyberjock, the data collected should never be wiped out, but then what about granularity? Is the data actually averaged out over time or is this only graphing artifact?