Tried to replace a drive in a raid 1 and all hell broke loose

[efkf@firemail.cc]

Hello, this is my first time interacting with a mailing list and the 
first time actually using btrfs.

THE CURRENT SITUATION:
I think I have two working drives with dm-crypt in what i think is raid 
1
when booted on a either fedora live disk or debian testing i can mount 
the filesystem but check fails and scrub aborts.
    -= FEDORA =-
The first time i scrub i get this on the kernel log: ( attachment 
fedora_scrub_1 )
i unmounted, mounted and scrubbed again and got a different error: ( 
attachment fedora_scrub_2 )
check says this: ( attachment fedora_check )
This is the distro info: ( attachment fedora_info ))

  -= DEBIAN TESTING =-
The first time i scrub i get this on the kernel log: ( attachment 
debian_scrub1 )
I unmounted and got a call trace and register dump on the kernel log: ( 
attachment debian_unmount )
i mounted and scrubbed again and got a similar error to fedora's second 
scrub: ( attachment debian_scrub2 )
check says the same thing
This is the distro info: ( attachment debian_info )

THE BACKSTORY:
here is a timeline of the events that lead up to this
(by the way, all of this was done with clean unmounts and reboots)
  [ Using debian stable btrfs-progs 5.10.1-2 ]
- Btrfs raid 1 works great srubs ok with 2 drives
- One drive starts dying, i get one to replace it
- Add the new one to the array (with all three plugged in) (i now know i 
should've use replace instead)
- run btrfs balance (with all three again)
- tried to delete the drive but failed (around this time i started 
running into weird errors)
- updated debian to testing, updated kernel, rebooted ( btrgs-progs 
5.10.1-2 -> 5.17-1 )
- managed to remove the failing drive (not sure if the update helped)
- Have to mount degraded and check shows a lot of errors
- bite the bullet and run check --repair, it says it is fixing errors 
and at the end it says no errors
- run just check again and get the same errors as before roughly
- run check --repair again, gets stuck in what seems to be an infinite 
loop. Says its fixing the same things again and again
- joined the irc channel to ask for help
- send sigterm to btrfs check --repair ( no one had replied by then)
- it actually mounts without needing -o degraded now
- get told to run 'btrfs fi us -T /mnt' and realize that i have both 
raid1 and some small single chunks
- run btrfs balance start -mconvert=raid1,soft -dconvert=raid1,soft /mnt
- btrfs check shows much less errors and all single chunks are gone ( 
the btrfs check output at that point is attached as old_completing_check 
on the email )
- try to run scrubs, they abort
- check and there are some single chunks yet again
- try fedora, run check and it gives the error i have attached and 
doesn't go through with actually checking everything
- try debian again and now checking doesn't complete with the same error 
as in fedora
I also tried running 'btrfs balance start -mconvert=raid1,soft 
-dconvert=raid1,soft /mnt' again now and the first time I run it, it 
says
ERROR: error during balancing '/mnt/sd': No such file or directory
and it the second time
ERROR: error during balancing '/mnt/sd': Read-only file system

Thanks a lot for reading my email and sorry if it winds up being 
something not related to btrfs