Weird Temp Behavior w/Seagate Enterprise Performance 1.2TB HDD 10K v7

gb00s

Active Member
Jul 25, 2018
228
61
28
Malta
Hi all,

Recently, I purchased a Supermicro SC216 chassis w/BPN-SAS2-216EL2 backplane. 1x LSI 9207-i8 HBA in IT-Mode is connected to the backplane. I installed the following disks and tested the drives for S.M.A.R.T and TEMP values.

12x Seagate ST1200MM0007 (1.2TB Enterprise Performance 10K v7)
10x HGST HUC109090CSS600 900GB 10K RPM SAS 2.5" HDD.

Before I start I should say that I exchanged the 3x original San Ace 80 fans with Noctua NF-A8 PWM (Chromax) and ARCTIC F8 PWM PST CO. Please do not judge the exchange as 'ridiculous' because of the much lower airflow, static pressure, or much lower rpm's. The 5EUR ARCTIC's are just doing fine and the following temp overview will show that an exchange has nothing to do with the 'weird issue'.

S.M.A.R.T values all without any issues. Super nice. But looking at the temps of all of the Seagate drives started to be a little bit 'worrisome'. All of the 12X Seagate drives showed temps between 60C and 74C just lying around and not doing any read/write jobs. The room temperature in the room was ~26C. The 10x HGST were installed beside the 12x Seagate drives. I also installed just the Seagates drives without the HGST drives. Same temps, no changes. I separated the 12x Seagate drives from each other with one drive slot only filled with caddies & dummies between them. I installed them in groups of 2 or 4 drives. I installed them all in the middle, all on the right, all on the left. The drives always showed never the same temps, but always between 60C and 74C. The HGST drives stayed all very cool between 35-43C. I think we can agree that the temps fro the Seagate drives were way to high and 10+C above hardcoded Trip Temperature of 50C. The backplane gave no warning at all. No beep, no LED indication that something was wrong. I also want to mention I tested with the San Ace 80 with 1-2C lower temps than the Noctua"s and ARCTIC's.

I measure the temps with the following command: smartctl -A /dev/sdx | grep -i temperature (x for each drive [a-s]).

Today I decided to test again, but only with 8x Seagate installed. With and without the HGST drives with surprising results. While the room temperature was 29C today, the 8x (!!!) Seagate showed temps between 49C and 55C only. Oooops. I repeated all the test from the day before with the same fans, the same formations of where the drives have been installed. I always had temps between 49C and 55C. How could that happen with 3C higher room temperature, same HBA, same backplane, same fans? The only difference was ... 8x vs 12x Seagate drives. So I tested it again with 12x of the Seagate drives and temps skyrocketed again to 64C and 77C in the peak!!!!!! Wxx ..... ??!! Switched back to 8x of the Seagate files and the temps slowly came down to 50-56C again. I did not test even less Seagate drives (6x or 4x). Needless to say the HGST drives stayed with the same temp all the time during the testing.

So my question is, how can the number of Seagate drives installed have such an impact on the temperature of the drives? As the HGST always kept their temperature it might have something to do with the Seagate drives or their firmware or ..... I don't know. Would be nice to hear some thoughts from you guys.

Thank you in advance.

Mike

PS: Just added a screenshot from the tests with 8x Seagate drives only. Forgot to make screenshots from 12x Seagate drives installed.

Temps_HGSTvsSeagate.png
 
Last edited:

Skud

Active Member
Jan 3, 2012
110
47
28
You shouldn’t be able to hold the drive in your hand for very long at those temps.
How do they feel?

Do you have an IR thermometer? Maybe leave a bay or two empty next to one of these drives and shoot the temperature that way.

Riley
 

gb00s

Active Member
Jul 25, 2018
228
61
28
Malta
How do they feel?
As you described. Uncomfortable to touch them ...
Maybe leave a bay or two empty next to one of these drives and shoot the temperature that way.
As described, I tested them side-by-side, with space (dummies) between them.

My question is why all(!!!) drives are becoming hot if 12 drives are installed, but temps with just 8 drives installed are close to normal? That's the thing I can't figure out. Will do more tests tonight.
 

Whaaat

Member
Jan 31, 2020
52
12
8
Will do more tests tonight.
Just a guess but HBA or drives themselves can initiate patrol read while idle, which is more power hungry than simply spinning hdd with offloaded ramp. What is the smartctl output regarding background scan?
 

gb00s

Active Member
Jul 25, 2018
228
61
28
Malta
Thanks for the comments.

In the meantime, this got weirder and weirder and a decent contact here already mentioned this sounds very like 'sailors yearn'. I can't deny it.

First of all the good news, with nice regulated room temperature ~24C I'm down to 37-41C in IDLE .... with 8x drives installed. With 12x drives installed still not so nice and still ~50C, which was the original 'Trip Temperatur' of the drives. So I found a 2015 update of the ST1200MM0007 firmware to revision 0003 directly from Seagate. I had no luck installing the DELL firmware upgrade from 2018. But I wanted to re-format the drives to 512 again before the firmware upgrade. To my surprise, the temperature of 8x installed drives immediately dropped down to 37-41C from above and from the ~50C I showed earlier in this thread. Temperatures with 12x disks installed dropped from the 60+C down to the ~50C on average. Ooopsi .... No firmware upgrade, just re-formating the drive. 'Sailors Yarn'. Then I upgraded all drives to revision 0003 firmware.

But the temperature difference between 8x and 12x drives installed still persists. Smaller, but it's still there. The new firmware increased the 'Trip Temperature' of each drive to 60C compared to 50C from revision 0002. My drives are around 50C in bigger than 8x disk installations ... :rolleyes: Hmmm ... Just a random change of 'Trip Temperature? I don't think so.

I remembered some drives had 'Non-Medium Error Counts' > 0 in the SMART reports before I bought these drives used. I didn't care as this value is in my opinion a 'controller'-issue (communication) only. Please, correct me if I'm wrong. I ran SMART-tests again today and realized that the 'Non-Medium Error Counts' values increase with 12x drives installed only. With just 8x drives installed this value doesn't change. This points to me to a controller or backplane issue. The question is, why a controller/backplane should act weird just with these Seagate ST1200MM0007? Firmware issue of the drives? Is there an issue with the extender chip? Is it getting to 'hot' with 12x drives and produces errors that lead to the higher power usage and therefore temps of the disks? The controller will be tested anyway, as I measured temp difference of 2C between disks connected to a DELL Perc and an LSI 9207-8i both in IT mode. Further tests will be:

1) Which HBA leads to these 'Non-Medium Error Counts' errors? ... and ...
2) Which drives are affected (I just mixed drives randomly until now).

What can I say?

Mike