blinking HDD-LED: SSD killer running or normal idle load?

Arnulf · 01-14-2022, 11:34 AM

HDD-LED blinks every two seconds while computer is running Slackware64 15.0 RC2 with Kernel 5.15.13 and any software is started by user action. In this case, normally a system should go into idle mode and doesn't perform any access to SATA controller.

dmesg output

"iostat", combined with "date", user does nothing between the two outputs shown below:

Code:

user@workstation:~$ date
Sat Jan  8 13:43:02 CET 2022
user@workstation:~$ iostat
Linux 5.15.13 (workstation)     01/08/22        _x86_64_        (16 CPU)

avg-cpu:  %user   %nice %system %iowait  %steal   %idle
           0.02    0.00    0.01    0.05    0.00   99.91

Device             tps    kB_read/s    kB_wrtn/s    kB_dscd/s    kB_read    kB_wrtn    kB_dscd
sda               4.78       167.67        12.43         0.00     619125      45897          0
sdb               1.10        65.36         0.00         0.00     241345          1          0
sdc               0.50         6.63         0.00         0.00      24483         12          0
sdd               0.42         4.78         0.01         0.00      17650         24          0
sr0               0.00         0.00         0.00         0.00          2          0          0


user@workstation:~$ date
Sat Jan  8 16:01:24 CET 2022
user@workstation:~$ iostat
Linux 5.15.13 (workstation)     01/08/22        _x86_64_        (16 CPU)

avg-cpu:  %user   %nice %system %iowait  %steal   %idle
           0.01    0.00    0.01    0.04    0.00   99.95

Device             tps    kB_read/s    kB_wrtn/s    kB_dscd/s    kB_read    kB_wrtn    kB_dscd
sda               1.67        51.62         5.70         0.00     619155      68333          0
sdb               3.70        23.25         0.00         0.00     278823          1          0
sdc               0.16         2.04         0.00         0.00      24493         12          0
sdd               0.13         1.47         0.00         0.00      17656         24          0
sr0               0.00         0.00         0.00         0.00          2          0          0

What causes 22436 KB ≈ 22 MB write load?

"iotop -o" only shows jbd2 process but jbd2 causes additional blinks of HDD-LED and not the blinks every two seconds.

Booting with "init 4", with "init 3" or with "init S" give nearly same results. Therefore X & GUI (KDE 5) can be excluded for this write load. Booting with unchanged delivered kernel 5.15.4 instead of kernel 5.15.13 gives nearly same result. Therefore a misconfiguration of kernel 5.15.13 can be excluded for this write load.

This write load doesn' t create any data except a few bytes written to syslog. Affected SSD isn't filled up.

Is this normal SATA idle load or is an SSD killer process still runnning?

jailbait · 01-15-2022, 08:21 PM

Quote:

Originally Posted by Arnulf

This write load doesn' t create any data except a few bytes written to syslog. Affected SSD isn't filled up.

What is written to syslog?

Arnulf · 01-16-2022, 08:02 AM

Only entries like followings are written:

/var/log/messages

Code:

Jan  8 14:41:37 workstation syslogd[895]: -- MARK --
Jan  8 14:41:48 workstation smartd[1363]: Device: /dev/sdb [SAT], SMART Usage Attribute: 194 Temperature_Celsius changed from 70 to 71 
Jan  8 15:11:48 workstation smartd[1363]: Device: /dev/sdb [SAT], SMART Usage Attribute: 194 Temperature_Celsius changed from 71 to 70 
Jan  8 15:41:37 workstation syslogd[895]: -- MARK --
Jan  8 15:41:58 workstation smartd[1363]: Device: /dev/sdd [SAT], SMART Usage Attribute: 190 Airflow_Temperature_Cel changed from 73 to 74 
Jan  8 15:41:58 workstation smartd[1363]: Device: /dev/sdd [SAT], SMART Usage Attribute: 194 Temperature_Celsius changed from 27 to 26

/var/log/syslog

Code:

Jan  8 14:42:34 workstation nmbd[1346]: [2022/01/08 14:42:34.924998,  0] ../../source3/nmbd/nmbd_browsesync.c:355(find_domain_master_name_query_fail) 
Jan  8 14:42:34 workstation nmbd[1346]:   find_domain_master_name_query_fail: 
Jan  8 14:42:34 workstation nmbd[1346]:   Unable to find the Domain Master Browser name HOME_100TX<1b> for the workgroup HOME_100TX. 
Jan  8 14:42:34 workstation nmbd[1346]:   Unable to sync browse lists in this workgroup. 
Jan  8 14:42:34 workstation nmbd[1346]: [2022/01/08 14:42:34.925053,  0] ../../source3/nmbd/nmbd_browsesync.c:355(find_domain_master_name_query_fail) 
Jan  8 14:42:34 workstation nmbd[1346]:   find_domain_master_name_query_fail: 
Jan  8 14:42:34 workstation nmbd[1346]:   Unable to find the Domain Master Browser name HOME_100TX<1b> for the workgroup HOME_100TX. 
Jan  8 14:42:34 workstation nmbd[1346]:   Unable to sync browse lists in this workgroup. 
Jan  8 14:57:34 workstation nmbd[1346]: [2022/01/08 14:57:34.935569,  0] ../../source3/nmbd/nmbd_browsesync.c:355(find_domain_master_name_query_fail) 
Jan  8 14:57:34 workstation nmbd[1346]:   find_domain_master_name_query_fail: 
Jan  8 14:57:34 workstation nmbd[1346]:   Unable to find the Domain Master Browser name HOME_100TX<1b> for the workgroup HOME_100TX. 
Jan  8 14:57:34 workstation nmbd[1346]:   Unable to sync browse lists in this workgroup. 
Jan  8 14:57:34 workstation nmbd[1346]: [2022/01/08 14:57:34.935626,  0] ../../source3/nmbd/nmbd_browsesync.c:355(find_domain_master_name_query_fail) 
Jan  8 14:57:34 workstation nmbd[1346]:   find_domain_master_name_query_fail: 
Jan  8 14:57:34 workstation nmbd[1346]:   Unable to find the Domain Master Browser name HOME_100TX<1b> for the workgroup HOME_100TX. 
Jan  8 14:57:34 workstation nmbd[1346]:   Unable to sync browse lists in this workgroup.

These entries are not written every two seconds.

Arnulf · 03-26-2023, 04:38 PM

Next try:
Different mass storage device configuration without any HDDs is now realized.
Slackware64 15.0 with Kernel 5.15.95 is running.
automount isn't active.

HDD-LED blinks every two seconds while computer is running independent of booting with "init 4", with "init 3" or with "init S".
Connected USB mass storage devices are "hit" at each HDD-LED blink.

Code:

user@workstation:~$ iostat 
Linux 5.15.95 (workstation)     03/26/2023      _x86_64_        (16 CPU) 

avg-cpu:  %user   %nice %system %iowait  %steal   %idle 
           0.18    0.00    0.09    0.04    0.00   99.69 

Device             tps    kB_read/s    kB_wrtn/s    kB_dscd/s    kB_read    kB_wrtn    kB_dscd 
nvme0n1           3.36        80.67        57.11         0.00    1234541     873996          0 
nvme1n1           0.09         2.96         0.00         0.00      45227          9          0 
sda               1.42         6.25         0.00         0.00      95673         19          0 
sdb               0.13         1.44         0.00         0.00      21991         61          0 
sr0               0.00         0.00         0.00         0.00          2          0          0

It looks like that a daemon checks every two seconds all connected mass storage devices and USB ports. Which daemon do this and how can it be stopped.

Slackware64 14.2 doesn't show a behavior described above, but this software is too old for current hardware. Therefore it isn't an option in this case.

lvm_ · 03-26-2023, 11:33 PM

iotop or KDE's ksysguard show real-time per-process disk activity

Arnulf · 03-27-2023, 11:00 AM

To omit any KDE5 stuff booted with init 3, iotop shows activity of jbd2 on /dev/nvme0n1p5 (Linux root partition, ext4) every 5 seconds. This doesn't correspond to HDD-LED blinkings every 2 seconds. These HDD-LED blinkings are annoying like a dropping water tap.

A hit to USB every two seconds is very bad for use of USB FDD because this may damage the FDD in a short time.

Jan K. · 03-27-2023, 01:34 PM

This 2-second issue has been reported here on LQ before with no solution... even asked the system monitor guys over at KDE how anything can run without being registered, but no one answered...

The're also reports of invisible, undetectable malware...

Arnulf · 03-27-2023, 04:43 PM

Quote:

Originally Posted by Jan K.

This 2-second issue has been reported here on LQ before with no solution...

Please post link(s) to related thread(s).

lvm_ · 03-28-2023, 04:21 AM

Disk LED on my notebook blinks just because it is made this silly way - even when there is 100% sustained load it blinks regularly instead of lighting all the time, so regular blinking may not be an indication of regular operation.