stats don't update during long external operation i.e. mkfs

[dankamongmen]

I sat for about five minutes watching an ext4 filesystem be created. During that time, growlight never indicated any stat updates. It would be interesting to see the feedback from these operations; ensure stats are able to update.

[dankamongmen]

I think the issue might actually be one with stats not collecting on this USB device. They don't seem to ever come on for the device.

[dankamongmen]

The USB device sees stats just fine. But yeah, I triggered a mkfs.ext4 from within growlight via asking for a new filesystem, and see no I/O, despite

1662882 be/4 root 0.00 B/s 0.00 B/s 0.00 % 99.99 % mkfs.ext~/dev/sdo1

and diskstats is increasing, though not with any great fury:

[schwarzgerat](0) $ cat /proc/diskstats  | grep sdo
   8     224 sdo 250 63 11960 585 1164 12482 2109996 460332 0 304148 460918 0 0 0 0 0 0
   8     225 sdo1 200 2 10552 153 1164 12482 2109996 460332 1 303924 460485 0 0 0 0 0 0
[schwarzgerat](0) $ cat /proc/diskstats  | grep sdo
   8     224 sdo 250 63 11960 585 1165 12489 2110004 462909 0 306724 463494 0 0 0 0 0 0
   8     225 sdo1 200 2 10552 153 1165 12489 2110004 462909 1 306500 463062 0 0 0 0 0 0
[schwarzgerat](0) $ cat /proc/diskstats  | grep sdo
   8     224 sdo 250 63 11960 585 1166 12496 2110012 465483 0 309300 466068 0 0 0 0 0 0
   8     225 sdo1 200 2 10552 153 1166 12496 2110012 465483 1 309076 465636 0 0 0 0 0 0
[schwarzgerat](0) $ 

but iotop shows the mkfs at 99.99% of the machine's I/O...though 0 actual bytes....hrmm.

[dankamongmen]

During an external mkfs.ext4, it is showing stats just fine:

╭──────[xhci_pci-0 [0] (256Gbps to chip, 133Mbps (0%) demanded)]───────────[-]─╮
│        sdo╭⇗⇨⇨⇨partition table metadata─────────────────────────────────────╮│
│           │m1111111111111111111111111 247.63G sdo1 1111111111111111111111111││
│      120Ki╰┤Survivor 3.0     000A 247.63G  512B gpt   n/a              PATA├╯│
╰─────[PCI Express 0000:05.00.3 (x16, gen 4.0)]────────────────────────────────╯

i was seeing 12Mi etc. during some of it. So I think we're just not updating stats while running one of these long-lived operations.

BTW, the mkfs.ext4 as we call it is just as slow outside as it is inside. Maybe that -b -2048 is killing us? Might want to figure this out. Writing the superblocks and filesystem accounting information takes just about forever.