Comment 5 for bug 1860182

Update: The scrub back on 4.15 went as expected. I rebooted with 5.3 kernel and tried the scrub again. I saw the same behaviour: the scrub started, with a very fast rate and short time estimate (3-5 minutes), and then reached 100% after about that time. However it continued to say it was in progress, despite saying 100.00% done. It did not calcualte a time estimate, saying the rate was slow, however, it did complete after 3h36 (so fast than it used to on kernel 4.15).

eric@eric-8700K:~$ zpool status
  pool: storagepool1
 state: ONLINE
  scan: scrub repaired 0B in 3h36m with 0 errors on Sat Jan 18 14:50:16 2020
config:

 NAME STATE READ WRITE CKSUM
 storagepool1 ONLINE 0 0 0
   mirror-0 ONLINE 0 0 0
     ata-WDC_WD20EZRZ-00Z5HB0_WD-WCC4M3YFRVJ3 ONLINE 0 0 0
     ata-ST2000DM001-1CH164_Z1E285A4 ONLINE 0 0 0
   mirror-1 ONLINE 0 0 0
     ata-WDC_WD20EZRZ-00Z5HB0_WD-WCC4M1DSASHD ONLINE 0 0 0
     ata-ST2000DM006-2DM164_Z4ZA3ENE ONLINE 0 0 0

errors: No known data errors

I started a scrub again to see whether it was just the first time with this kernel, but again the % complete seems inaccurate, and the time remaining is definitely inaccurate (0h0m shortly after starting).

So at the very least it is clear that zpool status is not accurately reporting the scrub process on kernel 5.3.