(cache)How to see how much data was read and written since boot?

exerceo · 09-15-2025, 02:43 PM

I would like to know how many GB of data were read and written by all processes since boot. Ideally I would like to see this statistic for each data storage device individually, but it would be good without this too. How can this be accomplished?

gnome-system-monitor shows this statistic on an individual process basis only.

teckk · 09-15-2025, 03:04 PM

Not sure that you can. Unless you kept tract of it at the start.

I suppose that you could look at Lifetime writes:

Example:

Code:

tune2fs -l /dev/sda2

Then look at it again later. Subtract the difference.

wpeckham · 09-15-2025, 03:12 PM

Set a process to report the total writes to storage early in the shutdown process, and then late in the reboot process, and have those log to a file. (Make sure logrotate is installed and configured to limit the growth of that log.) Now you can either go look at the log and do the math, or write a process to scan the log and do the math for you and report. That would give you the total written during shutdown and boot.

If your report program runs the same total writes detection and subtracts the logged number from the last boot you have the total written since boot.

If you get fancy you can create some neat graphs if you like.

I do not need this, but it sounds like a useful process for someone who NEEDS to know!

rnturn · 09-15-2025, 03:58 PM

Quote:

Originally Posted by exerceo

I would like to know how many GB of data were read and written by all processes since boot. Ideally I would like to see this statistic for each data storage device individually, but it would be good without this too. How can this be accomplished?

gnome-system-monitor shows this statistic on an individual process basis only.

See the "proc(5)" manpage and search for "diskstats" and the tip on what section of the kernel documentation to read next (probably /usr/src/linux/Documentation/admin-guide/iostats.rst). You might need to install the kernel sources package to see those files. Unfortunately, I see nothing that indicates the amount (MB, GB, whatever) of data that has been transferred to/from each device -- only the number of reads and writes (along with a bunch of timing data). If you're looking for some sort of performance indicator, reads and writes might help you locate overloaded disks (containing "hot" files).

Hope this helps a bit...

Cheers.

viel · 09-15-2025, 05:25 PM

Code:

[root@arcadia migration]# iostat
Linux 5.15.188 (arcadia.lo)     09/16/2025      _x86_64_        (8 CPU)

avg-cpu:  %user   %nice %system %iowait  %steal   %idle
           7.55    1.33    4.39    0.06    0.00   86.67

Device             tps    kB_read/s    kB_wrtn/s    kB_dscd/s    kB_read    kB_wrtn    kB_dscd
dm-0              0.38         4.28        23.84         0.00    2602255   14508612          0
sda               0.00         0.05         0.00         0.00      30118          0          0
sdb               0.50         6.29        31.09         0.00    3830503   18917864          0
sdc               0.00         0.06         0.00         0.00      33920          0          0


[root@arcadia migration]#

Maybe that helps.

GazL · 09-16-2025, 01:35 AM

Using --pretty and -dmN makes it look a little nicer. Especially so if, like me, you use device-mapper:

Code:

$ iostat --pretty -dmN
Linux 6.16.6-local (crux) 	09/16/25 	_x86_64_	(4 CPU)

      tps    MB_read/s    MB_wrtn/s    MB_dscd/s    MB_read    MB_wrtn    MB_dscd Device
     0.83         0.00         0.01         0.00        745       2130          0 lukssda4
     0.00         0.00         0.00         0.00          1          0          0 sysvg-lvswap0
     0.00         0.00         0.00         0.00          1          0          0 sysvg-lvslack150
     0.77         0.00         0.01         0.00        742       2130          0 sysvg-lvcrux38
     0.22         0.00         0.00         0.00        134        206          0 lukssda5
     0.02         0.00         0.00         0.00         17          0          0 datavg-lvslackware
     0.17         0.00         0.00         0.00        100        206          0 datavg-lvlocal
     0.01         0.00         0.00         0.00         10          0          0 datavg-lvlibrary
     0.00         0.00         0.00         0.00          4          0          0 datavg-lvwork
     0.00         0.00         0.00         0.00          1          0          0 datavg-lvmirrors
     0.53         0.01         0.01         0.00        889       2337          0 sda

mjolnir · 09-16-2025, 05:51 AM

Caveat - untested.

Code:

cat /proc/diskstats | awk '$6 > 0 || $10 > 0 {printf "%s - Read: %.2f GB, Written: %.2f GB\n", $3, $6 * 512 / 1073741824, $10 * 512 / 1073741824}'

syg00 · 09-16-2025, 07:03 AM

Conceptually ok, but what about 4k sector devices ?. And/or a mix of both.

The monitoring tools like iostat manage this for you - as the OP would have found out with a basic search. However they tend to be interval driven rather than a single period since boot, so the above would be also my preferred route.

viel · 09-16-2025, 10:06 AM

Well the two data seems come from the same font /sys/block/{disk}/stat

FWIW
https://www.kernel.org/doc/html/latest/block/stat.html