bad value display for "iostat -xdm", sysstat version:sysstat-12.5.5

[huiwenshu]

bad value display for "iostat -xdm", sysstat version:sysstat-12.5.5
When I use iostat -xdm 2 (or user sar -dm 2)to monitor. It always display bad value for me. Besides the bad value for "r/s" is a same value :9223372036854775808.00

[18:01:12]Device r/s rMB/s rrqm/s %rrqm r_await rareq-sz w/s wMB/s wrqm/s %wrqm w_await wareq-sz d/s dMB/s drqm/s %drqm d_await dareq-sz f/s f_await aqu-sz %util
[18:01:12]dm-0 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
[18:01:12]sdd 0.00 0.00 0.00 0.00 0.00 0.00 0.50 0.01 0.00 0.00 0.00 16.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.10
[18:01:12]sde 0.50 0.01 0.00 0.00 0.00 16.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.05
[18:01:12]sdf 9223372036854775808.00 4503599627370496.00 0.00 0.00 0.00 0.50 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 214748364.70
[18:01:12]sdg 1.00 0.02 0.00 0.00 0.50 16.25 0.50 0.00 0.00 0.00 2.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.05
[18:01:12]sdh 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
[18:01:12]sdi 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
[18:01:12]sdj 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
[18:01:12]sdk 0.50 0.00 0.00 0.00 1.00 0.50 0.50 0.00 0.00 0.00 0.00 0.50 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.15
[18:01:12]sdl 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
[18:01:12]sdm 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
[18:01:12]sdn 0.00 0.00 0.00 0.00 0.00 0.00 0.50 0.01 0.00 0.00 0.00 16.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.05
[18:01:12]sdo 1.00 0.01 0.00 0.00 0.50 8.25 0.50 0.00 0.00 0.00 0.00 0.50 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.15
[18:01:12]sdp 0.50 0.01 0.00 0.00 0.00 16.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.05
[18:01:12]sdq 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
[18:01:12]sdr 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
[18:01:12]sds 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
[18:01:12]sdt 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
[18:01:12]sdu 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
[18:01:12]sdv 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00

[sysstat]

It seems that there is a kernel counter which is decreasing with time although this shouldn't be possible.
To verify this, you can create a datafile with sar (e.g. sar -o datafile -d 2 ; stop it with ^C once the problem appeared) and send me the resulting datafile so that I can inspect its contents.

[huiwenshu]

sar.txt this is dump file,can you help me analyze the file ,thank you!

[sysstat]

I confirm: some kernel counters have decreased though this should never happen...
E.g. at 09:02:05 AM problem concerns /dev/sdc and /dev/sdn :

$ sar -f sar.txt -d -s 09:02:03 -e 09:02:05 --dev=sdc,sdn
Linux 5.4.17-2102.201.3.el7uek.x86_64 (cd-mesdb-rac01) 	02/21/2022 	_x86_64_	(104 CPU)

09:02:03 AM       DEV       tps     rkB/s     wkB/s     dkB/s   areq-sz    aqu-sz     await     %util
09:02:05 AM       sdc 9223372036854775808.00 4611686018427387904.00    315.75      0.00      0.00      0.00      0.00      0.25
09:02:05 AM       sdn      1.50 4611686018427387904.00     43.25      0.00     26.17      0.00      1.33      0.35

Here are the counters values collected by sadc from the kernel:

$ sadf -T -r -O debug sar.txt -- -d -s 09:02:03 -e 09:02:05 --dev=sdc,sdn
# name; A_DISK; nr_curr; 32; nr_alloc; 78; nr_ini; 78
09:02:05; major; 8; minor; 32; DEV; sdc; tps [DEC]; 45009321; 45009319; rkB/s [DEC]; 992424950; 992424872; wkB/s; 57354580; 57355843; dkB/s; 0; 0; rd_ticks [DEC]; 61036794; 61036793; wr_ticks; 1698078; 1698083; dc_ticks; 0; 0; tot_ticks; 8645440; 8645445; aqu-sz; 45851636; 45851637;
09:02:05; major; 8; minor; 208; DEV; sdn; tps; 3426592; 3426595; rkB/s [DEC]; 7954166; 7954150; wkB/s; 233575494; 233575667; dkB/s; 0; 0; rd_ticks; 164315; 164315; wr_ticks; 2438140; 2438144; dc_ticks; 0; 0; tot_ticks; 5858295; 5858302; aqu-sz; 757046; 757047;
# uptime_cs; 155628921; ust_time; 1645430527; extra_next; 0; record_type; 1; HH:MM:SS; 16:02:07

For /dev/sdc device, the total number of I/O (calculated as the sum of reads, writes and discards) goes from 45009321 to 45009319, and the total number of sectors read goes from 992424950 to 992424872.
For /dev/sdn, we get a total number of sectors read which goes from 7954166 to 7954150.
All those counters should be monotonic and always increase. I cannot tell you why this happens, but it looks like the problem lies in the kernel rather than in sysstat...