Seagate 8Tb Temperature

Status
Not open for further replies.

Mcfluffels

Cadet
Joined
Apr 5, 2018
Messages
5
Good Morning,
Here is some context.
About two days ago I bought a 8Tb Seagate Ironwolf (S# ZA1EDQRG).
First thing I did was Clone all of my Data from a 4Tb Seagate IronWolf, the pool was detached in freenas and and the drive unplugged and kept as an offline backup.
During the transfer the 8Tb Drive got to a maximum of 48 Degrees and stabilized there and the 4Tb maintained 35 Degrees during the 26 hour transfer.
When the 8Tb is not being busy with writes/reads it stabilizes at 44 Degrees, the 4Tb is in the drive caddy next to the 8Tb and is at 37 Degrees (Normally at 27 Degrees)
Both Drives are in a 5-in-3 drive cage with a Corsair SP120 120mm fan, there are no airflow issues

My Questions?
- Is it normal for the 8Tb to run at such a high temperatures. (Note it is stabilized at 44 Degrees and fluctuates +/- 2 Degrees) Compared to the 4Tb that stays around 27 Degrees during no/moderate usage.
- I have heard that using advanced power management to spin down the drive to lower temperature results in more wear and tear compared to keeping it always on because of the large changes in temperature between idle/usage. Is this true?
- Will the high temps affect drive longevity?
 

SweetAndLow

Sweet'NASty
Joined
Nov 6, 2013
Messages
6,421
Is it a 7200rpm drive or 5400rpm? 7200 usually run hotter. Mid 40's are probably ok but you do have an air flow problem even if you don't think you do. Turn that fan speed up. My fans run at 4000rpm and drives still hit 40c in the summer months.
 

Mcfluffels

Cadet
Joined
Apr 5, 2018
Messages
5
It is a 7200rpm drive, I will turn up the fan speed when i get home see if it helps.
 

SweetAndLow

Sweet'NASty
Joined
Nov 6, 2013
Messages
6,421
It is a 7200rpm drive, I will turn up the fan speed when i get home see if it helps.
Yeah that's just what 7200 rpm drives do. Idle around 40c with that is fine.
 

Chris Moore

Hall of Famer
Joined
May 2, 2015
Messages
10,080
- I have heard that using advanced power management to spin down the drive to lower temperature results in more wear and tear compared to keeping it always on because of the large changes in temperature between idle/usage. Is this true?
Yes
 

silverback

Contributor
Joined
Jun 26, 2016
Messages
134
I had to raise my alarm temperature when I moved from Seagate 4tb to 8 TB. It is now set at 45c. Probably to high for some.
 

Chris Moore

Hall of Famer
Joined
May 2, 2015
Messages
10,080
I like to see disk temperatures in the 30s, but there are only a few ways to get there with 7200 RPM disks. Either have the room really cold, uncomfortably cold by my standards, or have the fans in the chassis running pretty fast.
Some combination of those two. It's a sliding scale. It is because of the heat factor that I prefer to use the lower RPM drives in my system at home but I do have a few of the faster drives and they stick out like a sore thumb in my status reports because of the temperatures which are (on average) in my system at home, ten degrees higher. I have 12 drives that are all at 31 and four that are at 42 and all the hot ones are the fast ones. There are some other drives in there also with a range of temperatures below 42, but you get the idea.
 

Constantin

Vampire Pig
Joined
May 19, 2017
Messages
1,828
Apologies in advance but please allow me to quibble a bit.
I like to see disk temperatures in the 30s, but there are only a few ways to get there with 7200 RPM disks. Either have the room really cold, uncomfortably cold by my standards, or have the fans in the chassis running pretty fast.
Some combination of those two. It's a sliding scale.
I'd argue that temperatures depend in large part on the drive and the environment they are mounted in. Some drives run cooler, some run hotter by default. FarmerPlings guide certainly should provide a good insight into the thermal load of each drive as the thermal load should scale with the electrical load while operating. There may be subtle cooling performance differences between drives based on shapes and materials, but I doubt they are material.

For example, my 10TB 7,200 RPM He10 spinners from HGST normally run at 30-31*C and their fans are too quiet for me to hear. I use a dedicated fan controller and three 120mm fans that blow across twelve 3.5" slots. The A76 case provides ample distance between each drive and lots of air slots in the mounting hardware. This is one of the reasons I like this case even though it is gargantuan and a pain to move around. I did add some duct tape to shape air flow across the drives, however.

In the past, I used HGST 3TB, 7,200 RPM drives in the same enclosure that ran considerably hotter. Even within that cohort, there were some drives that were noticeably hotter than others. See the IR picture below. All the drives are similar HGST 3TB drives, yet one is noticeably hotter than the rest.
20180113t203257-jpg.22354

However, I was able to keep them around 35*C or less with the above case, fan controller, and fans. The backplane combined with the adequate room between hard drives is the main reason I like Lian Li so much. I'm OK with trading off some hard drive density in return for quiet operation.

What currently makes my rig noisy is trying to keep the D-1537 and LSI 2216 SAS chips cool. I use two industrial 120mm fans which keep both the CPU and the SAS adequately cooled (sub 35*C for the CPU, sub-50*C for the SAS). I plan on improving the cooling of both chips via dedicated blowers with shaped volutes and all that. Some have also had success replacing the heat sink on the Xeon D chips with coolJag and like coolers. No such luck for the SAS chip though, as its heat sink appears to be a unicorn (measurement-wise) and its location is problematic (right in line with a PCIe Slot, yay!).

The best hard drive toaster I encountered thus far is the Norco RPC-431, with the Mini XL taking second place. While the Norco RPC-431is ludicrously bad re: air flow for the disk drives, the upper hot swap bays in the Mini XL do not fare much better. My 3TB HGSTs hit 50*C in these during scrubs. I blame the single-minded focus on minimizing case fans inside the Mini XL. A second 80mm+ fan higher up would have helped considerably.
 
Last edited:

pro lamer

Guru
Joined
Feb 16, 2018
Messages
626
8Tb to run at such a high temperatures. (Note it is stabilized at 44 Degrees and fluctuates +/- 2 Degrees) Compared to the 4Tb
You may also try exchanging these two's bays and learn what happens...

Just a wild idea.

Sent from my phone
 

Ericloewe

Server Wrangler
Moderator
Joined
Feb 15, 2014
Messages
20,175
8 Terabits? That's a pretty unusual way of saying 1 Terabyte.

Silliness aside, nomenclature and correct use of units matters. 1 TB = 8 Tb
 

Chris Moore

Hall of Famer
Joined
May 2, 2015
Messages
10,080
I'd argue that temperatures depend in large part on the drive and the environment they are mounted in. Some drives run cooler, some run hotter by default.
That isn't an argument, it is saying the same thing in a different way.
I have a 48 bay server like this as my primary FreeNAS system at home:


1556653077533.png click to enlarge...
The drives on the front row, right behind the inlet fans stay between 28°C and 31°C. The next row of drives is a couple of degrees hotter and the back row of drives is usually around 36°C to as much as 38°C. Most of the drives (21 of 32, if I recall correctly) are Seagate Desktop drives (ST4000DM000) like this:

1556653445263.png click to enlarge...
These 5900 RPM drives run cooler that the three Seagate Constellation drives that are 7200 RPM drives. They are in the same pool with the Desktop dives. The Constellation drives look a little different, so you can tell the design is different, but I think the big deciding factor on the heat is the RPM. I have the Constellation drives mounted in the front row to help keep them cool, where the other drives are in the low 30s, the Constellation drives are in the high 30s but not usually over 40°C, still it is almost ten degrees difference, while mounted in the same row of the same chassis, with the same air flow, same room, same capacity, same drive vendor... I only see one difference to justify the temperature difference, the RPM.
Also in this chassis, I have a pool made up of Western Digital Gold 6TB drives, like this:
1556654063643.png click to enlarge...
Those are also 7200 RPM drives and they are the hottest of all. They are all running between 38°C and 42°C very consistently ten degrees hotter than the slower spinning Seagate Desktop drives. I don't have a fancy IR camera to look at but I do monitor my system closely and I have been doing so since the first system I built back in 2011 with used drives I bought from eBay. I have (over the years) had drives from every vendor, Samsung Spinpoint for example, in every size from 1TB up and this is just speaking of the ones I have at home.

I also do this for work where I have servers under management with drives from Seagate, HGST, and Western Digital. The most recent server we acquired was one like this:
1556654580000.png click to enlarge...
This chassis is filled with the Seagate Exos 10TB helium filled drives. They run (in our uncomfortably cold data-center) under 30°C all the time with the coolest drives being around 26°C, but I can't say if that is the fault of the chassis having great airflow, the drives being cool running drives because of the helium despite their 7200 RPM spin speed, or if it is because the room is cold at 68°F ambient.
The thing I can say is that we have 6 disk shelves full of 4 TB WD Red Pro drives in the same rack and those drives all run in the mid 40s despite the room temperature and the airflow from the disk chassis.
I think that the helium filled drives run cooler in general.
I also think that airflow matters a lot, because I have tinkered with the airflow in my personal rack chassis systems at home.
I also am pretty certain that the RPM of the drive matters. It even appears to make a difference what the ambient temperature in the room is. All these things must be taken into account, but the airflow in the chassis is probably the easy one to control.
See the IR picture below. All the drives are similar HGST 3TB drives, yet one is noticeably hotter than the rest.
Cool thermal photo.
 
Last edited:

Constantin

Vampire Pig
Joined
May 19, 2017
Messages
1,828
My A76 case just blanched in envy. Some day, I may yet buy a used Supermicro CSE 836, mount it sideways on a wall and be able to use all 16 SAS2 / SATA3 ports for a 2-VDEV Z3 array. :)

But the drive cooling in the A76 is pretty awesome. I'm going to build a shelf for it in the meantime and then a 2nd one for the console monitor and keyboard.
 
Last edited:

Frikkn

Cadet
Joined
Nov 12, 2018
Messages
6
I have been reading up on this the past few days because I have a pair of drives that run at ~45C. I was surprised to find out a lot of HDDs have a limit of 60C which I found positively surprising. I would have no problem running a CPU 15C below the maximum temperature so why would it be different for HDDs? What that implys to me is that simply approaching the maximum operating temperature should not increase failure rate by any significant means, until you go above it.
There is plenty of data out there about temperature vs HDD life, like Backblazes for example. The lack of correlation has eased me up a bit about drives running at 40C+. This still requires you to read up on the hard drives specs, as they will differ. For example WD Reds are rated at up to 60C, while Seagate Ironwolfs are rated at "Operating (drive case, max °C) 70".
 

Chris Moore

Hall of Famer
Joined
May 2, 2015
Messages
10,080
I was surprised to find out a lot of HDDs have a limit of 60C which I found positively surprising.
Hard drives can endure higher temperatures, but they last longer if you keep them cool. Around a year ago, we had a series of posts on the forum from someone that lives in a desert environment where they are not able to air-condition the storage system and it continuously runs over temp. Their drives last around a year to a year and a half before needing to be replaced. High failure rate due to heat. Between 30°C and 40°C is a good place for long lasting dries, but if you must run your drives hotter, monitor them closely because they will fail more frequently.
like Backblazes for example.
Not a good example. There is a correlation.
 

Constantin

Vampire Pig
Joined
May 19, 2017
Messages
1,828
HDDs are not solid state devices... Don’t expect the bearing grease to be as happy / unhappy about high temperatures as the silicon inside a CPU. Ditto capacitors, esp. if they’re the electrolytic kind.

I aim to keep my motherboards, CPUs and drives reasonably cool without reverting to exotica like Peltier devices.
 

Chris Moore

Hall of Famer
Joined
May 2, 2015
Messages
10,080
you would look at real data
I have real data of my own. If you would have looked at some of the more than eight thousand posts I have made you might have seen some of it.
n the meantime for anyone that is interested in actual data here is another study presented at Usenix FAST 16 and best paper award winner
https://www.usenix.org/system/files/conference/fast16/fast16-papers-manousakis.pdf
Authors: Ioannis Manousakis, Rutgers University; Sriram Sankar, GoDaddy; Gregg McKnight, Microsoft; Thu D. Nguyen, Rutgers University; Ricardo Bianchini, Microsoft
Another interesting conclusions that points towards humidity being a bigger factor in drive failure than temperature or temperature variations.
The problem with the data that you cite is that it is only applicable in their specific conditions. Did you actually look at it? The reason they have high humidity is because they are not using air conditioning in the normal way because they are trying to be more cost effective with regard to electricity. In a normal air conditioned data-center or a normal home environment, the humidity is controlled by the air conditioning. The hardware they are using and data-center design is not representative of most small business data facilities and it is certainly not representative of the home environment where most people keep their NAS. The data you have presented is not relevant.
 
Last edited:

Constantin

Vampire Pig
Joined
May 19, 2017
Messages
1,828
I don't doubt that the data may be hazy on this - drive reliability in the face of higher or lower usage patterns, temperature, humidity, and any number of other factors may be affected in non-consistent manners. Unless the whole data set is available, it's hard to say how representative the conclusions are. Backblaze has not made it available, AFAIK.

I will note that despite the conclusions that Backblaze is drawing, their data centers are not open-air systems. They run their drives well within the allowable band of temperatures so I would not draw conclusions on drive reliability if you want to run your drives at much higher temperatures. Let's take a look at the Backblaze report in question (see here).

Little to no data is published. We learn that the lowest drive temperature is 12*C, the highest is around 38*C. For the drive families they pulled out for comparison sake, the temperature range was from 16-28*C for Seagate, and 21-31*C for Hitachi. I'd argue that few US homes feature ambient air temperatures around 53*F, which is the maximum it can be at Backblaze. Likely, it's lower because a drive will be warmer than the air being pulled across it. In my home, the delta is about 6*C for an A76 case with it's excellent air flow. So,
  • Would I draw conclusions about running drives in the 40*C+ range based on data collected in the mid-20's (on average)? No.
  • Do I observe Backblaze still heavily cooling their facility despite this research? Yes.
  • Is the presented Backblaze hard drive life vs. temperature data representative of long-term reliability? Unlikely. Most of their drives are replaced due to obsolescence rather than hardware failure. (Hello Deathstar!)
  • Please also consider the closing sentence of that web page: "As long as you run drives well within their allowed range of operating temperatures, keeping them cooler doesn’t matter." [emphasis mine]
In other words, go ahead and "cook" your drives, if you like. I'd expect the failure rate during the warranty period to remain somewhat within actuarial limits (after all, most external enclosures feature terrible heat management). How much longer the drives will last after the warranty period expires is a different question, however.

Alternatively, consider how the online backup industry spends billions of USD cooling their drives annually, presumably for a reason. If it really didn't matter, these drive farms would be operating at much higher temperatures than they do. Some providers (Apple, MS, Facebook, LinkedIn, etc.) are building server farm hardware with higher allowable temperature limits for precisely that reason.

In this report, Backblaze didn't intentionally segregate their storage pods into various climatic zones to compare lifetime at 10*C, 20*C, 30*C, 40*C, 50*C, and 60*C ambient air temperature. Until someone does run a large controlled experiment of this sort and subsequently does not find a significant correlation re: drive life, I'd stick to industry standards, especially for rotating drives.
 
Last edited:

Chris Moore

Hall of Famer
Joined
May 2, 2015
Messages
10,080
And another paper by Microsoft and University of Virginia which is also a good read
https://cheetah.cs.virginia.edu/~gurumurthi/papers/acmtos13.pdf
Authors: SRIRAM SANKAR, Microsoft Corporation MARK SHAW, Microsoft Corporation KUSHAGRA VAID, Microsoft Corporation SUDHANVA GURUMURTHI, University of Virginia
Here is a quote from the paper you cited:
Yang et al [1999] establishes that a 25 C delta in temperature derates the MTTF by a factor of 2 in their study on Quantum hard disk drives. Cole et al [2000] from Seagate, present thermal de-rating models showing that MTTF could degrade by close to 50% when going from operating temperatures of 30C to 42C. Our results agree with the observations made by Cole. Our measured failure rates also exceed the AFR rates that manufacturers mention in their datasheets
I guess you didn't actually read it because what that says is that going from 30C to 42C increases failure rate by 50%
 
Last edited:

Constantin

Vampire Pig
Joined
May 19, 2017
Messages
1,828
Well, that underscores again why I'm happy with my spinners operating at 31*C. The embedded CPU is not far behind at 34*C.

The SAS controller chip (LSI 2116) on the motherboard really likes to run hot, however. I have a hard time keeping it below 50*C, and that bothers me because the chip is part of the motherboard.
 

Chris Moore

Hall of Famer
Joined
May 2, 2015
Messages
10,080
Don't worry I never thought for even a second that you would look at real data, hence why I said "In the meantime for anyone that is interested in actual data".
I initially didn't want to look at the data you presented, but I am pretty pleased that I did. The last paper you cited, "the paper by Microsoft and University of Virginia (you claimed it was, "also a good read") https://cheetah.cs.virginia.edu/~gurumurthi/papers/acmtos13.pdf
Authors: SRIRAM SANKAR, Microsoft Corporation MARK SHAW, Microsoft Corporation KUSHAGRA VAID, Microsoft Corporation SUDHANVA GURUMURTHI, University of Virginia" , gives us this additional quote that is nice to see as it is quite similar to my own experience:
The result of our study is surprising since earlier studies [Pinheiro et al. 2007] establish that disk drive failures do not increase with increase in temperature in the field. Figure 5 shows the actual HDD temperature in increments of one degree and the corresponding AFR for our entire population. We see clearly that with increase in HDD temperature, the AFR rate increases. There are some data points at the end of the spectrum that have smaller number of samples and hence a higher skew. For the major part of the distribution, we see that AFR steadily increases as HDD temperature increases.
1556849406102.png
The thing I have observed over the last decade or so is that hard drives have become a bit more reliable, but the temperatures do still affect the rate of failure.
 
Status
Not open for further replies.
Top