# DF-3215 V1 attempts — DEFAULT-config (512MB UNDO/REDO FIFO) race, stock kernel #0.
# Captured from guest /tmp/df3215/v1_p*.log + v1b.log before reboot (tmpfs).
#
# Attempt 1 (4 parallel unprivileged writers x 16 files x 40 rounds x 1MB):
===> v1_p0.log <==
round 40/40: 640 MB of redo records laid
done, no panic: total redo laid ~640 MB
===> v1_p1.log <==
round 40/40: 640 MB of redo records laid
done, no panic: total redo laid ~640 MB
===> v1_p2.log <==
round 40/40: 640 MB of redo records laid
done, no panic: total redo laid ~640 MB
===> v1_p3.log <==
round 40/40: 640 MB of redo records laid
done, no panic: total redo laid ~640 MB
# => 2.56 GB of REDO laid into the 512MB FIFO, no panic (QEMU virtio flusher
#    kept up).  vfs.hammer.stats_redo after: 3071956232.
#
# Attempt 2 (+4GB of plain dirty data to stall finalize reclaim):
round 40/40: 640 MB of redo records laid
done, no panic: total redo laid ~640 MB
trigger_rc=0
# vfs.hammer.stats_redo after: 3840020720 (3.84GB cumulative).
# => DEFAULT-config overflow NOT reproduced on this guest (page-cached
#    virtio storage reclaims too fast).  Claim limited to small-undo
#    filesystems (demonstrated at 8MB) — plausible but unproven on
#    slower real storage.