FreeNAS 9.2.1-RC2 (2nd Release Candidate) now available for download

[jkh]

The problem with nagios is that it just satisfies one kind of reporting system - then you have folks asking for integration with others, since they don't use nagios, and before you know it FreeNAS has suffered an explosion of reporting mechanisms. That is why we went to the new alert system with a simple REST API interface for querying its status - folks can plug that into whatever they like and we don't bear the burden of keeping the other reporting systems up to date. In any case, I still stand by my assertion that IPMI is the way to go if you want hardware telemetry information and a nice reporting interface.

 

[cyberjock]

jkh hit it right on the head. We've had many people ask about reporting systems. People often want stuff that is obscure, closed sourced, or not even available on FreeBSD. The discussion always ends with "the best way to monitor your hardware is with OS agnostic reporting tools. IPMI is just such a tool. You just can't come up with a valid argument against IPMI. As for other monitoring(such as hard disk monoring) FreeNAS does have it's own SMART monitoring and will email you if there's a problem(assuming you've setup FreeNAS appropriately). But between IPMI and FreeNAS, you have a total monitoring solution that should cover all of the bases.

 

[Magnus33]

Where were you expecting to get that information? A lot of systems don't have a separate service processor that is wired into all of the sensor data (and those motherboards which do have some array of sensors, don't necessarily support getting that information in the same way, which makes a generic solution somewhat difficult). This is really what IPMI is all about - it both collects and exports this data in a convenient way, also logging events. If your hardware is not IPMI capable, you're Doing It Wrong and this isn't really a flaw in FreeNAS so much as a flaw in your hardware.

And yes, iXsystems sells systems with IPMI. Very soon, even the lowest-end product will have it, and it's got a lot of great sensor data.

Understandably there are limits to hardware at present tell the new systems roll out but shouldn't the most common hardware be supported at this point?
If you don't have one or two of the most common hardware then your out of luck but if you do very basic info like temperature reports can go a long way to diagnosing failing hardware.

If the guys in the forum can slap a script together to do this for the most common hardware its should be easy enough to include this in freenass (i use the term easy loosely :) )

No one is expect support for all hardware solutions but for those of us who do buy the common hardware and monitor nas's from a distance this can be a god send.
Its a lot easier to say slap out this harddrive when you leave the office when you see a large spike in temp and aviod a harddrive failure during business hours when you can actually see them temperature in real time.

 

[zambanini]

i hope ix will sell ipmi and other extensions as plugins.

 

[jgreco]

IPMI is a hardware feature of your motherboard. It cannot be added through software, though software can be used to gather at least some of the data IPMI can give you.

However, most attempts to do it with software run into various problems; the standards are poor and/or poorly implemented, and there are several of them.

IPMI has the singular benefit of being a "standard" way to do it, out of band even.

 

[Magnus33]

What he said.
Although the other side of this is many of the freenas users are not going to have ipmi boards which is why i figure freenas should have some kind of reporting for the most common method they use.

Obscure or closed sourced and you out of luck and really it can't be expected.

 

[jgreco]

It is hard to do, well, portably. Some of us have decades of experience... crap gear that locks up when you try to do X, or lies about collected data 'cuz the mfr "doesn't support that"... etc

 

[jkh]

Understandably there are limits to hardware at present tell the new systems roll out but shouldn't the most common hardware be supported at this point?

If that hardware does not support IPMI, then no. It's too "common", in the other sense of that word. :)

Hell, there is a metric buttload of 32 bit Intel hardware out there too. Probably far more of it, if you look at the entire world's population, than there is 64 bit Intel hardware since a lot of the old stuff that the first world does not want goes to 3rd world countries. That will not stop us from killing 32 bit support in 9.2.2 (seriously, 9.2.1 is the last one for 32 bit!). Sometimes "common" is so far from "good" that you have to draw the line!

On the bright side, if you look at some of the newer motherboards (even in Mini-ITX), you'll see that IPMI is getting more prevalent. As I said before, soon even iX's lowest-end hardware will offer it.

 

[DJ9]

On the bright side, if you look at some of the newer motherboards (even in Mini-ITX), you'll see that IPMI is getting more prevalent. As I said before, soon even iX's lowest-end hardware will offer it.

Should be interesting, since I know you guys have something brewing. ;)

 

[Magnus33]

Here were a problem begins when your counting on hardware to fill the role that software does not.
It changes so rapidly that the solution your counting on either gets tossed aside or doesn't last as long as you like.

Plex actually ran into this problem recently when it changed plex android to depend greatly on android video playback but never bother to take into account that even minor changes to android os can break plex playback.
With so many different makers android tends to get modded to the device its on and plex ran into huge problems.

There a good reason ipmi has started showing on cheaper products.

The question is the next gen solution going to be better for everyone or to expensive for the laymen.

Fortunately there always coders coming up with ideas and outside hardware monitoring solutions.