Re: Re: patch fr1-2.17 does not work with kernel 2.4.30

"Also sprach Stephan von Krawczynski:"
> Ok, I see. This is only recognisable if you feed the speed_limit_min|max with
> MAXINT. That's quite ok so far.

Well, it does whatever the original code would have done, given the
real figures on i/o speed :-). It's another question if that is sensible
or not. I've suggested that you try taking out the if
(md_need_resched(current)) block before the speed calculation, to see
empirically if it helps (or hinders).

> Another thing I noticed is that it seems impossible to hotadd to a raid with
> persistent superblock with bitmap support.

Not sure ...

> If you do the test case from the FR1 readme (mkraid -> raidsetfaulty -> raidhotadd) you get:

This is preciely when HOTREPAIR occurs. I tested that myself here. It
works.

> kernel: md: syncing RAID array md0
> kernel: md: minimum _guaranteed_ reconstruction speed: 2147483647 KB/sec/disc.
> kernel: md: using maximum available idle IO bandwith (but not more than 2147483647 KB/sec) for reconstruction.
> kernel: md: using 124k window, over a total of 25679808 blocks.
> kernel: md0: removed bitmap e605bc80

Why? Has it finished already? I'd need to see more of the messages to
know.

> and following a complete resync.

??

> Is this intended behaviour?

No - quite the opposite occurs, and is intended to occur! I posted the
test:

  sh-2.03# mdadm -C -l 1 -n 2 -x 0 /dev/md/0 /dev/ubd/[12]
  mdadm: array /dev/md/0 started.
  sh-2.03# cat /proc/mdstat
  Personalities : [raid1]
  read_ahead 1024 sectors
  md0 : active raid1 ubd/disc2/disc[1] ubd/disc1/disc[0]
      4032 blocks [2/2] [UU]

  unused devices: <none>
  sh-2.03# sync
  sh-2.03# raidsetfaulty /dev/md/0 /dev/ubd/2
  sh-2.03# echo hi > /dev/md/0
  sh-2.03# sync
  sh-2.03# raidhotadd /dev/md/0 /dev/ubd/2
  sh-2.03#

  md: syncing RAID array md0
  md: minimum  guaranteed  reconstruction speed: 100 KB/sec/disc.
  md: using maximum available idle IO bandwith (but not more than 100000
    KB/sec) for reconstruction.
  md: using 124k window, over a total of 4032 blocks.
  md0: retained bitmap a027dc60
  raid1: synced dirty sectors 0-7

  # raid1: skipped clean sectors 8-8063
  md: md0: sync done.


All I can imagine is that the event count did not match, but I'd need
to see more of the kernel messages to know.


Peter