Concerned about temperature

[broadcast]

Hey all, just installed unRAID after getting rid of my Thecus 2100 NAS device, one thing I noticed was lack of a temperature triggered shutdown, my Thecus had a module that would shut the device down if the drive temps reached a certain threshold or if the fan failed.  I did a search of the forums here but couldn't come up with much, except that it seems there is really no automatic way of shutting down the server...  but in the meantime, is there any quick way (sorry I'm a total linux noob) of setting up a cron job that calls up a script that checks drive temp and if drive temp is above a threshold to execute a shutdown? 

 

Thanks!

[RobJ]

There is no reason I can think of why we don't have a script for that already, sounds very doable.  I suspect one or more of our 'script masters' is working on one now ...  Currently at least 3 of the UnRAID Add Ons display in red any temp that is too high, but I don't believe any of them act on it yet.  Probably coming ...

[bubbaQ]

Spundown drives don't' return temps (except for some recent exceptions).  Now you can assume that a spundown drive won't overheat, but that's not always true.

 

And (until the latest beta) hardware monitoring was not enabled in the kernel, so you could not test system temps at all.

 

These are not reasons to not do the script... it certainly would work on drives that are spun up... but recognize the limitations.  I have a script that does this, in addition to controlling the fan speeds in my system.  Runs every minute via cron.

 

Also, what reaction strategy to your employ?  Spindown and unmount the one hot drive, and let the array continue? (high availability).  Or shutdown the system?  If you do the former,  what do you do with a second overheating drive?  What about parity?

[NAS]

The reaction strategy is key and eberyone will likely want something differernt.

 

However its probably a good idea to have the system beep.

 

Obviously that would require us requesting system beep be compiled into the kernel again (which im going to do now as it has alot of potential use)

[SSD]

Hey all, just installed unRAID after getting rid of my Thecus 2100 NAS device, one thing I noticed was lack of a temperature triggered shutdown, my Thecus had a module that would shut the device down if the drive temps reached a certain threshold or if the fan failed.  I did a search of the forums here but couldn't come up with much, except that it seems there is really no automatic way of shutting down the server...   but in the meantime, is there any quick way (sorry I'm a total linux noob) of setting up a cron job that calls up a script that checks drive temp and if drive temp is above a threshold to execute a shutdown? 

 

Great idea!  I've been a long time advocate of something like this.  See this link.  Having email go out at less extreme levels is a great idea.

 

Spundown drives don't' return temps (except for some recent exceptions).  Now you can assume that a spundown drive won't overheat, but that's not always true.

 

All modern WD drives (500G+) report temps while spun down going back into 2006, and maybe even earlier (hard for me to call that a "recent exception" ).  That is 1/2 of my drives.  More likely the problem is that the monitoring tool is unaware that the WD drives CAN report temperature while spun down and therefore don't ask.  myMain and the new custom GUI do this, but the default GUI and smarthistory don't have these smarts. 

 

But point is well taken - if a Seagate drive is spun down and the drive gets hot, it will not report a high temperature. 

 

Hard to imaging why that would happen, though.  I feel you'd be pretty well protected against things like HVAC failure with hardware monitoring.  Having one WD drive would also do the trick.  At a finer grain, having one WD drive in each "bank" of drives would allow a user to be notified if a fan went out.

[bubbaQ]

notified if a fan went out.

 

Tom has enabled hardware monitoring in the kernel in the latest beta.  Coupled with some other tools like lmsensors and dmidecode, I plan to do a monitoring app that will monitor drives, as well as system temps and fanspeeds.  Even can control fan speeds if the mobo supports it.

 

Just have to wail till the Web/php flavor of unRAID comes out.

 

 

[olympia]

http://lime-technology.com/forum/index.php?topic=2470.0

 

This mod has this functionality for months now, however I never tested it.

[gaikokujinkyofusho]

Hi, I thought I'd first try to revived this discussion before startign something new as it seems that some people either had some things going or thought something should be out there.

 

I have a server packed with drives and have just about given up on keeping it 100% cool, I am sure its a combination of things (like too many drives [10]), not enough fans (five 80mm, one 140mm), ... though mostly just not well organized (air mostly comes in through the top vent and is blown over the drives and  out everywhere else). Any my old place doesn't have AC

 

Anyway. Not sure my tower is going to get much cooler. I would rather it shut down/sleep, or a disk sping down, when they get too hot rather than have constant access to it. I have it set to send me email alerts but can't seem to find any other options (v6.1.9) regarding temperature, eps actions to take when it gets too hot.

 

Are there any options that I am missing, or perhaps plugins/dockers that I perhaps missed that would add this kind of fucntionality? This would really be helpful.

 

Thanks!