UFS Find the process that increases the temperature of the disk.

dmtr

dmtr

Good day to all!

The question is a bit strange.
But he is.

I use smartd (smartmontools).
Sometimes, I get warnings about the high temperature of the disk (Like a mail message).

The question is that I can't figure out which process does it: raises the temperature.
This happens every time at a different time.

How to catch a pest?
 
It would help to understand how hot the disks are getting, and what sort of disks (make and model).

It's possible that some application is exercising your disks, but in a sensibly configured system, your disks should not over-heat, no matter how hard they are worked.

Overheating your disks can cause serious damage. Generally spinning disks should not exceed 60C and SSDs might be able to go to 70C (but check the specs for your specific drives). Exceeding the specified operating temperature will void the warranty on some drives (the better ones retain temperature history).

You should examine the options to improve the cooling inside your case. Consider flow direction of existing fans (whether they push or pull, in concert or against each other), more fans, better fans, better fan positioning (directly blowing cool air onto all disks), and maybe a better case (designed for superior cooling).
 
Even if you push a disk with 100% busy I/O it still shouldn't "overheat". Having a really active disk does warm it up of course, but never to the extend it starts to overheat. I've had SATA cards where the controller chip would get hot enough to burn your fingers. Needless to say, that card was faulty. The disk itself could be broken, causing the electronics to overheat. Or you have really bad airflow in your cabinet, causing heat to build up, which in turn triggers the overheat alert.
 
Thanks for the answer and recommendations.

My thresholds are at the lowest: the limit of 42 degrees Celsius.
And it may seem strange :)
And the normal operating mode is 33-36 degrees Celsius.

But the question is, what process takes up the disk so much that the disk temperature increases so much?

Alternatively: write a script to analyze the temperature and the list of processes.
As soon as the temperature exceeds the limit, remember the list of processes.
 
Code: 
Model Family:     Hitachi/HGST Travelstar Z7K500
Device Model:     HGST HTS725050A7E630
194 Temperature_Celsius     0x0002   162   162   000    Old_age   Always       -       37 (Min/Max 19/61)
 
HGST (formerly IBM) has a generally superior reputation for making quality disk drives.
There is no indication of exceeding operating temperature limits in what you post above.
Please explain what you mean by "the limit of 42 degrees Celsius". In particular, how is this limit determined/set?
Please show us the output of:
Code: 
grep "^DEVICESCAN" /usr/local/etc/smartd.conf
Please show us the temperature history. Assuming your disk is /dev/ada0:
Code: 
smartctl --log=scttemphist /dev/ada0
You are still barking up the wrong tree thinking that any application needs to be identified or blamed for making your disk hot.
If your disk is too hot, then it is either broken, or it is insufficiently cooled.
 
gpw928 said:
HGST (formerly IBM) has a generally superior reputation for making quality disk drives.
There is no indication of exceeding operating temperature limits in what you post above.
Please explain what you mean by "the limit of 42 degrees Celsius". In particular, how is this limit determined/set?
Please show us the output of:
Code: 
grep "^DEVICESCAN" /usr/local/etc/smartd.conf
Click to expand...
I am set own limit .. as it seems to me, normal.

Code: 
/usr/local/etc/smartd.conf
/dev/ada0 -a -o on -S on -I 194 -W 10,38,42 -R 5 -m root

Maybe it's paranoia :)

gpw928 said:
Please show us the temperature history. Assuming your disk is /dev/ada0:
Code: 
smartctl --log=scttemphist /dev/ada0
Click to expand...
Code: 
=== START OF READ SMART DATA SECTION ===
SCT Temperature History Version:     2
Temperature Sampling Period:         1 minute
Temperature Logging Interval:        1 minute
Min/Max recommended Temperature:      0/60 Celsius
Min/Max Temperature Limit:           -40/65 Celsius
Temperature History Size (Index):    128 (70)

Index    Estimated Time   Temperature Celsius
  71    2022-04-13 08:02    39  ********************
  72    2022-04-13 08:03    39  ********************
  73    2022-04-13 08:04    38  *******************
  74    2022-04-13 08:05    38  *******************
  75    2022-04-13 08:06    39  ********************
  76    2022-04-13 08:07    38  *******************
 ...    ..(  5 skipped).    ..  *******************
  82    2022-04-13 08:13    38  *******************
  83    2022-04-13 08:14    39  ********************
  84    2022-04-13 08:15    38  *******************
  85    2022-04-13 08:16    38  *******************
  86    2022-04-13 08:17    39  ********************
  87    2022-04-13 08:18    38  *******************
 ...    ..( 18 skipped).    ..  *******************
 106    2022-04-13 08:37    38  *******************
 107    2022-04-13 08:38    37  ******************
 108    2022-04-13 08:39    37  ******************
 109    2022-04-13 08:40    38  *******************
 ...    ..(  2 skipped).    ..  *******************
 112    2022-04-13 08:43    38  *******************
 113    2022-04-13 08:44    37  ******************
 ...    ..( 17 skipped).    ..  ******************
   3    2022-04-13 09:02    37  ******************
   4    2022-04-13 09:03    38  *******************
   5    2022-04-13 09:04    37  ******************
 ...    ..( 64 skipped).    ..  ******************
  70    2022-04-13 10:09    37  ******************

gpw928 said:
You are still barking up the wrong tree thinking that any application needs to be identified or blamed for making your disk hot.
If your disk is too hot, then it is either broken, or it is insufficiently cooled.
Click to expand...

You're probably right.
After all, the temperature rise occurs rarely and for a short time.
 
dmtr said:
I am set own limit .. as it seems to me, normal.
Click to expand...
No, 42C is well within the "normal" range.
dmtr said:
Maybe it's paranoia :)
Click to expand...
Yep. Your disk is not overheating. The operating temperature range specified by HGST for your disk is 0C to 60C.
Now, you don't really want to get anywhere close to 60C, but 42C is a long way off that limit.
Ignoring attribute 194 ("-I 194") in smartd.conf(5) is a worry. You should review that!
For my disks, rated for 0C to 60C, I use:
Code: 
DEVICESCAN -a -o on -S on -n standby,q -s (S/../.././02|L/../../6/03) -W 3,50,55 -R 5!  -m diskmaster@mailhost
 
42°C ain't hot. For a technical system that's slightly warmed up. Naturally because there is energy put in to make the device operate.
(With 1.8W the HGST HTS725050A7E630 is very low in power consumption for a 7200rpm HDD. Other HDDs are easily above 5..8W or even more.)
If you touch it and pull your finger back while saying "Ouch!" - that's hot ? That would be above 60°C


What always should be kept in mind is in technical systems we are always dealing with two kinds of units:
metric system and the archaic UK/US feet & fathoms crap.
(I just love this Length Units graph. It shows angloamerican units do not needed to be compared with any other unit system, a closer look on itself discloses how ridiculous obsolete it is per se.)

However:
It may also be a possibilty that the delivered value internally measured by the disk is in Fahrenheit, while all the rest is handling it as it was Celsius.
Celsius vs. Fahrenheit shows roomtemperature is already above your alarm treshold of 60° if Celsius and Fahrenheit are confused.

(After all, on lowest hardware level there are only numbers. It's software that declares: "This (binary) value is this or has that unit."
There was an interesting article about units in code linked on HN in march.)
 
Thx.
Confusing units may not be the cause of your problem, but experience in electronics teached me one things or the other, such as:
"Damned! Why do those drill holes in the PCB not match the ones of the housing?!"
"Well, it may be feet."
"wha..????"
"By default many EDAs are configured working in mil/inch, while your CAD works in meters."

There are many funny and not so funny stories (an Ariane crash was caused by that) about that confusing units problem.
That's why I said: Always keep that in mind - it's quickly overseen.
 
top(1) (the page for 14.0-CURRENT is easier to read)

An I/O mode example:

Code: 
% top -m io -n 7 -d 4 -s 5
last pid: 11779;  load averages:  1.03,  1.82,  1.62; battery: 99%  up 0+10:16:19    18:31:38
184 processes: 1 running, 183 sleeping
CPU: 11.5% user,  0.3% nice,  9.5% system,  0.1% interrupt, 78.5% idle
Mem: 3640M Active, 5216M Inact, 2355M Laundry, 4154M Wired, 360M Free
ARC: 2003M Total, 1160M MFU, 121M MRU, 4775K Anon, 157M Header, 561M Other
     683M Compressed, 1813M Uncompressed, 2.66:1 Ratio
Swap: 16G Total, 832M Used, 15G Free, 5% Inuse

  PID USERNAME     VCSW  IVCSW   READ  WRITE  FAULT  TOTAL PERCENT COMMAND
19592 grahamperr566316529 8341080 601299 839295 287659 1728253  48.17% firefox
11247 grahamperr 39791  16731      4     18      4     26   0.00% firefox
22023 grahamperr10817616 3553110   3525  15637   8319  27481   0.77% firefox
11771 root       13440   1119  13298      0    163  13461   0.38% git
 9940 grahamperr 48316   9731      0      0      0      0   0.00% firefox
 2866 root      6301442 14107571   1090   1692 477067 479849  13.38% Xorg
 3089 grahamperr3595119 349486   9348   1293 141873 152514   4.25% plasmashell

last pid: 11789;  load averages:  0.94,  1.79,  1.61; battery: 99%  up 0+10:16:24    18:31:43
187 processes: 1 running, 185 sleeping, 1 zombie
CPU:  3.1% user,  0.0% nice,  5.9% system,  0.3% interrupt, 90.6% idle
Mem: 3661M Active, 5242M Inact, 2363M Laundry, 4086M Wired, 365M Free
ARC: 2010M Total, 1170M MFU, 116M MRU, 16M Anon, 158M Header, 550M Other
     697M Compressed, 1789M Uncompressed, 2.57:1 Ratio
Swap: 16G Total, 832M Used, 15G Free, 5% Inuse

  PID USERNAME     VCSW  IVCSW   READ  WRITE  FAULT  TOTAL PERCENT COMMAND
11783 root         366     18    422      0    388    810  39.88% git
19592 grahamperr  5377    151      0     23      0     23   1.13% firefox
11771 root         786    311    771    124     46    941  46.33% git
 9940 grahamperr   568     68      0      0      0      0   0.00% firefox
22023 grahamperr   906    102      0      0      0      0   0.00% firefox
 3186 grahamperr    11      1      0      0      0      0   0.00% gkrellm
 2787 netdata      287      6      0      0      0      0   0.00% netdata

last pid: 11793;  load averages:  1.43,  1.87,  1.64; battery: 99%  up 0+10:16:29    18:31:48
186 processes: 1 running, 185 sleeping
CPU:  3.1% user,  0.0% nice, 85.1% system,  0.1% interrupt, 11.7% idle
Mem: 3661M Active, 5274M Inact, 2363M Laundry, 4085M Wired, 343M Free
ARC: 1997M Total, 1185M MFU, 112M MRU, 1100K Anon, 158M Header, 542M Other
     704M Compressed, 1774M Uncompressed, 2.52:1 Ratio
Swap: 16G Total, 832M Used, 15G Free, 5% Inuse, 40K In

  PID USERNAME     VCSW  IVCSW   READ  WRITE  FAULT  TOTAL PERCENT COMMAND
11792 root         265    826    299      0    461    760  68.22% git
19592 grahamperr  2777    389      6     23      4     33   2.96% firefox
22023 grahamperr   577    228      0      0      0      0   0.00% firefox
 3389 grahamperr   250    109      0      0      0      0   0.00% konsole
 3186 grahamperr     8     27      0      0      0      0   0.00% gkrellm
 2866 root         229    275      0      0    154    154  13.82% Xorg
 2787 netdata      152     75      3      9      0     12   1.08% netdata

last pid: 11807;  load averages:  1.31,  1.84,  1.63; battery: 99%  up 0+10:16:34    18:31:53
184 processes: 2 running, 182 sleeping
CPU:  7.0% user,  0.1% nice,  4.8% system,  0.2% interrupt, 88.0% idle
Mem: 3592M Active, 5273M Inact, 2363M Laundry, 4092M Wired, 407M Free
ARC: 2003M Total, 1175M MFU, 107M MRU, 22M Anon, 158M Header, 541M Other
     693M Compressed, 1760M Uncompressed, 2.54:1 Ratio
Swap: 16G Total, 832M Used, 15G Free, 5% Inuse

  PID USERNAME     VCSW  IVCSW   READ  WRITE  FAULT  TOTAL PERCENT COMMAND
19592 grahamperr  5430     76      7     31      0     38   3.65% firefox
 3089 grahamperr  2578    204      5      3    208    216  20.75% plasmashell
 2866 root        2085    838      0      0     45     45   4.32% Xorg
11060 grahamperr   112     28      0      0      0      0   0.00% firefox
22023 grahamperr   864     89      0      0      0      0   0.00% firefox
 3186 grahamperr    17      5      0      0      0      0   0.00% gkrellm
 2787 netdata      283     10      0      0      0      0   0.00% netdata

%
 
You must log in or register to reply here.
Back
Top