That's sorta what I expected..... First thing I'd try -- turn off UMA for ZFS. It will have a performance impact but it eliminates one of the issues that I had to do a lot of work to get around in the patch I developed.
In /boot/loader.conf: vfs.zfs.zio.use_uma=0 And reboot. On 12/11/2015 01:56, Alexander Leidinger wrote: > On Thu, 10 Dec 2015 07:29:03 -0600 > Karl Denninger <[email protected]> wrote: > >> On 12/10/2015 05:03, Alexander Leidinger wrote: >>> On Wed, 9 Dec 2015 15:15:10 -0600 >>> Dustin Wenz <[email protected]> wrote: >>> >>>> Are you by chance using the ARC patch from this PR? >>>> >>>> https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=187594 >>> No (despite the fact that I would like to have it hit the tree). >>> >>>> Do you have a vfs.zfs.dynamic_write_buffer tunable defined, if so, >>>> what is it set to? >>> No. >>> >>> Bye, >>> Alexander. >>> >> Wait a minute... you have a panic coming out of running the periodic >> jobs and you do _*not*_ have the patch loaded? Got a dump available >> from that? >> > To make it explicit about what is patched on this system or not: > # svnlite status > M etc/rc.d/localpkg > M lib/libc/Makefile > M lib/libutil/Makefile > M lib/msun/Makefile > M libexec/rtld-elf/Makefile > M sbin/dump/dumprmt.c > M sbin/dump/itime.c > M sbin/dump/optr.c > M sbin/dump/tape.c > M sbin/restore/dirs.c > M sbin/restore/symtab.c > M sbin/restore/tape.c > ? sys/amd64/conf/ANDROMEDA > M sys/dev/drm/drmP.h > M sys/dev/e1000/if_igb.c > M sys/dev/usb/serial/ulpt.c > M sys/fs/cd9660/cd9660_vfsops.c > M sys/fs/msdosfs/msdosfs_vfsops.c > M sys/fs/smbfs/smbfs_vfsops.c > M sys/fs/udf/udf_vfsops.c > M sys/fs/unionfs/union_vfsops.c > M sys/gnu/fs/reiserfs/reiserfs_vfsops.c > M sys/kern/kern_jail.c > M sys/kern/kern_sig.c > M sys/kern/subr_bus.c > M sys/sys/jail.h > M usr.sbin/adduser/adduser.sh > M usr.sbin/adduser/rmuser.sh > M usr.sbin/nfsd/nfsd.c > M usr.sbin/watchdogd/Makefile > > Unfortunately I don't have a dump. Every crash so far didn't lead to a > core in the dump device (the thread on stable has a stacktrace: > https://lists.freebsd.org/pipermail/freebsd-stable/2015-December/083807.html). > No idea why I don't get a dump. > > # grep dump /etc/rc.conf /boot/loader.conf > /etc/rc.conf:dumpdev="/dev/ada3p2" > /boot/loader.conf:dumpdev="/dev/ada3p2" > > # savecore -C /dev/ada3p2 > > # ll /var/crash > total 9 > -rw-r--r-- 1 root wheel 3B Nov 11 08:45 bounds > -rw-r--r-- 1 root wheel 5B Apr 28 2011 minfree > > Bye, > Alexander. > -- Karl Denninger [email protected] <mailto:[email protected]> /The Market Ticker/ /[S/MIME encrypted email preferred]/
smime.p7s
Description: S/MIME Cryptographic Signature
