On Apr 28, 2011, at 3:56 PM, Jeremy Chadwick wrote: > On Thu, Apr 28, 2011 at 11:33:22PM +0930, Daniel O'Connor wrote: >> Does anyone else use ZFS to store TM backups? >> >> I find that whenever my laptop (over wifi!) starts a TM the ZFS machine it's >> backing up to grinds to a halt.. Other systems streaming stuff over NFS from >> it also tend to stall.. >> >> I presume that TM is doing something which causes ZFS some issues but I'm >> not sure how to find out what the real problem is let alone how to fix it.. >> >> I am running FreeBSD midget.dons.net.au 8.2-PRERELEASE FreeBSD >> 8.2-PRERELEASE #8 r217094M: Sat Jan 8 11:15:07 CST 2011 >> dar...@midget.dons.net.au:/usr/obj/usr/src/sys/MIDGET amd64 >> >> It is a 5 disk RAIDZ1 with 1.29Tb free using WD10EADS drives. >> >> I don't see any SMART errors or ZFS warnings. >> >> I have the following ZFS related tunables >> >> vfs.zfs.arc_max="3072M" >> vfs.zfs.prefetch_disable="1" >> vfs.zfs.txg.timeout=5 >> vfs.zfs.cache_flush_disable=1 > > Are the last two actually *working* in /boot/loader.conf? Can you > verify by looking at them via sysctl? AFAIK they shouldn't work, since > they lack double-quotes around the values. Parsing errors are supposed > to throw you back to the loader prompt. See loader.conf(5) for the > syntax.
In my /boot/loader.conf I have: vfs.zfs.arc_max=6200000000 vfs.zfs.vdev.min_pending=4 vfs.zfs.vdev.max_pending=8 vfs.zfs.txg.timeout=5 vfs.zfs.prefetch_disable=1 And they all are properly reflected in sysctl values (no parse errors seen). > Next, you COULD try using Samba/CIFS on the FreeBSD box to see if you > can narrow the issue down to bad NFS performance. Please see this post > of mine about tuning Samba on FreeBSD (backed by ZFS) to get extremely > good performance. Many people responded and said their performance > drastically improved (you can see the thread yourself). The trick is > AIO. You can ignore the part about setting vm.kmem_size in loader.conf; > that advice is now old/deprecated (does not pertain to you given the > date of your kernel), and vfs.zfs.txg.write_limit_override is something > you shouldn't mess with unless absolutely needed to leave it default: > > http://lists.freebsd.org/pipermail/freebsd-stable/2011-February/061642.html Just wanted to note that we were having a terrible time with ZFS performance. Copying just a single large file over the network would bring interactive system usage to an absolute crawl (system is 2x xeons, 12gb ram). Thanks to your optimizations we have great ZFS + Samba performance. We also use Netatalk for afpd file sharing (though I have not tried Netatalk as a Time Machine target) and our performance is quite good there as well. Scott _______________________________________________ freebsd-stable@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-stable To unsubscribe, send any mail to "freebsd-stable-unsubscr...@freebsd.org"
