Date: Sun, 27 Sep 2009 11:43:49 +0200 From: "O. Hartmann" <ohartman@mail.zedat.fu-berlin.de> To: Olivier Smedts <olivier@gid0.org> Cc: freebsd-current@freebsd.org, krad <kraduk@googlemail.com>, Stefan Bethke <stb@lassitu.de>, Dmitry Morozovsky <marck@rinet.ru> Subject: Re: ZFS on entire disks setup Message-ID: <4ABF33D5.2060909@mail.zedat.fu-berlin.de> In-Reply-To: <367b2c980909261132n60bd0789s1e3077c8ac6eab0b@mail.gmail.com> References: <alpine.BSF.2.00.0909261534060.62389@woozle.rinet.ru> <6A973ECE-83E7-4653-BBE7-CC3093361D19@lassitu.de> <d36406630909261111t75c3aa99y4606cb2c7758406b@mail.gmail.com> <367b2c980909261132n60bd0789s1e3077c8ac6eab0b@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Olivier Smedts wrote: > 2009/9/26 krad <kraduk@googlemail.com>: >> 2009/9/26 Stefan Bethke <stb@lassitu.de> >> >>> Am 26.09.2009 um 13:39 schrieb Dmitry Morozovsky: >>> >>> =EF=BF=BDDear colleagues, >>>> is there a way to configure ZFS-only setup without partitions? >>>> >>>> I tried to reproduce the trick with `skip=3D1 seek=3D1024', and have= >>>> loot/loader >>>> running, but it does not see the pool. >>>> >>> This sequence is working for me: >>> dd if=3D/boot/zfsboot of=3D/dev/da0 count=3D1 >>> dd if=3D/boot/zfsboot of=3D/dev/da0 skip=3D1 seek=3D1024 >>> zpool create zroot /dev/da0 >>> zpool set bootfs=3Dzroot zroot >>> cd /usr/src && make installworld installkernel distribution DESTDIR=3D= /zroot >>> cp /boot/loader.conf /zroot/boot/loader.conf >>> cp /etc/rc.conf /zroot/etc/ >>> touch /zroot/etc/fstab >>> echo 'zfs_load=3D"YES"' >>/zroot/boot/loader.conf >>> echo 'vfs.root.mountfrom=3D"zfs:zroot"' >>/zroot/boot/loader.conf >>> zpool export zroot >>> zpool import zroot >>> cp /boot/zfs/zpool.cache /zroot/boot/zfs/ >>> zfs set mountpoint=3Dlegacy zroot >>> >>> ... except that it doesn't anymore. =EF=BF=BDI saw this working about= two months >>> ago, but now it fails to mount root. =EF=BF=BDVerbose boot doesn't gi= ve any >>> indication why ZFS can't find the pool. >>> >>> >>> Stefan >>> >>> -- >>> Stefan Bethke <stb@lassitu.de> =EF=BF=BD Fon +49 151 14070811 >>> >>> >>> >>> >>> >>> _______________________________________________ >>> freebsd-current@freebsd.org mailing list >>> http://lists.freebsd.org/mailman/listinfo/freebsd-current >>> To unsubscribe, send any mail to "freebsd-current-unsubscribe@freebsd= =2Eorg" >>> >> >> have the performance issues with zvols been fixed. I remember a while = ago it >> wasn't a good idea to run swap on a zvol. This is why i left a partiti= on >> free for swap >=20 > Still not a good idea, I've got lockups for many minutes when > consuming heavy memory ressources, for example when doing a "make" in > ports/emulators/virtualbox with firefox and amarok opened. I've got > 2GB RAM on amd64. >=20 >> _______________________________________________ >> freebsd-current@freebsd.org mailing list >> http://lists.freebsd.org/mailman/listinfo/freebsd-current >> To unsubscribe, send any mail to "freebsd-current-unsubscribe@freebsd.= org" >> >=20 I see those lockups on three boxes when compiling or having heavy disk I/O on three different boxes, one (private) UP amd64 with 2GB, one SMP 4-core box amd64 with 8 GB and one SMP 8-core box amd64 SMP with 16 GB RAM. It doesn't matter what filesystem is involved, either UFS2 or ZFS! When compiling world AND compiling KDE4-libs or GCC44/GCC45, even those boxes with 8 or 16 GB of RAM and 4 or 8 cores have painful performance issues (locking for several seconds, no response over network or, if enabled and running, X11/mouse gets jumpy and completely inresponsible, quite near useless that moment). I doubt that ZFS causes the problem, it sounds more like a scheduling problem introduced with the early 8.0 (since 7.2 ran fluently on mentioned boxes, but have had other issues). All three boxes under my supervision run FreeBSD 8.0-RC1/amd64 and they all use UFS2-partioned harddrives for OS separated from data drives using ZFS-only. Even copying/backup data from /home on a ZFS volume to another ZFS volume with compression enabled (on another SATA-II drive) does not reveale those issues on the 2GB, UP amd64 box as showed when the box is under heavy load when the GCC compiler is running or PERL scripts are doing their duties. Regards, Oliver
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4ABF33D5.2060909>