Again, constant disk activity from Python on fresh install...disks never spin down

TrueNAS General
, ,
1

Good morning,

I know this topic has been covered many times but maybe never solved.

Fresh install
Moved system dataset to boot pool
No jails, VMs, nothing running
No GUI
Tried rebooting many times
SMART off
Dedub off

There is constant and equal pattern of writing to the disks, the noise they produce is always the same (same pattern) and the IO GFX shows the same pattern every few seconds.

It looks like Python keeps generating disk IO, here is the result by using most CPU time:

322  -  S      1:20.15 python3.9: middlewared (python3.9)
 324  -  I      0:00.15 /usr/local/bin/python3.9 -c from multiprocessing.resource_tracker import main;main(14)
 457  -  I      0:23.12 /usr/local/bin/python3.9 -c from multiprocessing.spawn import spawn_main; spawn_main(tracker_fd=15, pip
 494  -  I      0:01.86 /usr/local/bin/python3.9 -c from multiprocessing.spawn import spawn_main; spawn_main(tracker_fd=15, pip
 610  -  I      0:18.59 /usr/local/bin/python3.9 -c from multiprocessing.spawn import spawn_main; spawn_main(tracker_fd=15, pip
 611  -  I      0:19.51 /usr/local/bin/python3.9 -c from multiprocessing.spawn import spawn_main; spawn_main(tracker_fd=15, pip
 654  -  I      0:18.12 /usr/local/bin/python3.9 -c from multiprocessing.spawn import spawn_main; spawn_main(tracker_fd=15, pip
 655  -  I      0:17.11 /usr/local/bin/python3.9 -c from multiprocessing.spawn import spawn_main; spawn_main(tracker_fd=15, pip
2069  -  I      0:00.40 python3 /usr/local/bin/wsdd.py (python3.9)
1625 v0  Is+    0:00.99 python3 /etc/netcli (python3.9)

I love TrueNAS, best FreeBSD solution. Not an hardcore NAS user hence this constant disk activity (and energy consumption), will move me away from it…

Thanks for any help, appreciated,

Fabio

2

And you have tried spinning down the drives? Disks not spinning down | TrueNAS Community

3

Hi, thanks for the reply. This is my disk setup (2 same disk in mirror):

HDD Standby: 5 min
APM: Minimum power usage (spindown)
Force HDD standby: YES

Screenshot 2024-04-20 at 11.13.36
Screenshot 2024-04-20 at 11.13.36954Ă—536 14.4 KB

I will try the settings of the article now and let you know,

Thank you

F

4

BTW, as you can see, SMART has been disabled hence the options in the article are not related to my case…I guess :slight_smile:

5

I would actually enable the smart service and try to set everything as it should, with a greater than 5 minutes spindows timer.

6

Nothing to do…I also tried implementing the spindown script depicted here: GitHub - ngandrass/truenas-spindown-timer: Monitors drive I/O and forces HDD spindown after a given idle period. Resistant to S.M.A.R.T. reads.

but nothing…disk keeps getting accessed and written to by that Python process…

I truly do not know how to solve this…

7

That’s not normal behaviour. Do you have any cronjob or int scripts that are located on the HDDs?

8

No init scripts, only a couple of rsync tasks that get executed every 24 hours.

I will try to disable them?

9

Done, no change…disks activity is still there.

I also checked all the logs and there is nothing special in any of them

It is all about that python process…

Thanks anyway for the help.

10

I will look into this and come back tomorrow with hopefully something.

11

Thank you so much, please let me know if there is anything you may need from me…

12

Could you please tell me more about your hardware? Also, htop might help.

13

It is:
ASRock J3710
16Gb RAM
2 x 2TB WD Drive
32Gb USB Stick for OS

Is that enough?

14

Which model? Might be the SMR thing doing SMR things.

15

Excuse me, that was another NAS.

The disks for this are:

2 x 2TB SEAGATE ST2000LM003

16

Are these 2.5’’ HDDs? Then they are most likely SMR and should be avoided at all costs.

17

Correct, they are 2.5 HDDs…what is wrong with them?

Just checked, they are PMR

18

Here Is some info covering that topic. SMR vs CMR

19

Those are not SMR, they are PMR…

And I am still wondering what that has to do with that Python process who keeps stealing IO cycles :slight_smile:

20

As far as i know SMR is the shingled version of PMR. And CMR is not.

So if your drives are shingeld, the problem might be "SMR doing SMR stuff.