Balancing does not help with "no space left" despite capacity available.

No luck. I copied off my 10TB and am ready to delete, but after a failed balance it now is stuck in read only mode. Even after reboot it goes into read only mode as soon as I try to delete the first additional file. This started after I tried 0,10,20,30 (0 chunks) and finally 40 found 3 chunks, but it failed. Before that I was able to delete files OK (notice i have an extra 1tb now, and a couple of gig of metadata).

Seems like I have a brief window of rw after a reboot, and that ro is triggered by the first file delete.

root@ReadyNas:~# btrfs filesystem df /nas
Data, single: total=54.07TiB, used=51.73TiB
System, DUP: total=8.00MiB, used=6.09MiB
Metadata, RAID1: total=449.88MiB, used=447.72MiB
Metadata, DUP: total=244.50GiB, used=242.56GiB
GlobalReserve, single: total=512.00MiB, used=0.00B

root@ReadyNas:~# mount | grep /nas
/dev/md127 on /nas type btrfs (ro,noatime,nodiratime,nospace_cache,subvolid=5,subvol=/)
/dev/md127 on /run/nfs4/nas/Data type btrfs (ro,noatime,nodiratime,nospace_cache,subvolid=263,subvol=/Data)

the logs:

root@ReadyNas:~# dmesg | tail -n 50
[ 298.049409] BTRFS: error (device md126) in __btrfs_free_extent:7004: errno=-28 No space left
[ 298.049411] BTRFS info (device md126): forced readonly
[ 298.049412] BTRFS: error (device md126) in btrfs_run_delayed_refs:2995: errno=-28 No space left