Crash during Hammer overnight cleanup
I had a crash overnight. It looks like it happened during batch cleanup on a busy Hammer disk. This system has a separate PFS, /build, for doing bulk build work. There's a ton of activity on the disk because of that bulk build, so I created that PFS and upped the cleaning times earlier in the day previous to this crash. The bulk build was running at the time.
Dump at: leaf.dragonflybsd.org:~justin/20121004-dump.tgz (200M) Upload will complete about half an hour after this ticket is submitted.
DragonFly 3.3-DEVELOPMENT DragonFly 181ede8-DEVELOPMENT #0: Wed Oct 24 15:19:15 MSD 2012 firstname.lastname@example.org:/usr/obj/home/justin/32/src/sys/X86_64_GENERIC x86_64
(The bulk build was happening in a 3.2 chroot.)
The /build hammer config:
snapshots 0d 0d
prune 1d 50m
rebalance 1d 50m
dedup 1d 50m
reblock 1d 50m
recopy 30d 10m
Here's /var/log/messages from the relevant timeframe:
Nov 4 03:03:50 kernel: Warning: busy page 0xffffffe000b5f5a0 found in cache
Nov 4 03:05:27 kernel: HAMMER debug: shifted cursor pointing at parent
Nov 4 03:05:27 kernel: parent 8000000c7409e000:22 onode 8000000c7cf21000:0 nnode 8000000c6e14e000:20
Nov 4 03:13:14 kernel: Warning: busy page 0xffffffe000aac428 found in cache
Nov 4 03:13:14 kernel: Warning: busy page 0xffffffe003009ce8 found in cache
Nov 4 03:14:01 kernel:
Nov 4 03:15:47 syslogd: kernel boot file is /boot/kernel/kernel
Nov 4 03:15:47 kernel: Copyright (c) 2003-2012 The DragonFly Project.
Nov 4 03:15:47 kernel: Copyright (c) 1992-2003 The FreeBSD Project.