Date: Fri, 21 Nov 2008 10:37:04 +0200 From: Nikolay Denev <ndenev@gmail.com> To: freebsd-current@freebsd.org Subject: Re: HEADS UP: New ZFS in the tree. Message-ID: <29D6BF8D-6A05-4446-95FE-8F0AC0195C79@gmail.com> In-Reply-To: <18E318DD-29FA-4E26-89CF-11B893B42E34@gmail.com> References: <20081117205526.GC1733@garage.freebsd.pl> <8ECD400F-BFE3-4E31-94F0-39AF5F44FDAC@gmail.com> <20081119090307.GA81236@icarus.home.lan> <alpine.BSF.2.00.0811190323110.59299@borg> <FD98E80A-30DC-4700-9C0F-0D77835DBB32@gmail.com> <18E318DD-29FA-4E26-89CF-11B893B42E34@gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On 19 Nov, 2008, at 13:58 , Nikolay Denev wrote: > -----BEGIN PGP SIGNED MESSAGE----- > Hash: SHA1 > > > On 19 Nov, 2008, at 11:48 , Nikolay Denev wrote: >> >> Well, it looks like that on -current and a 4G amd64 machine >> probably there is no need >> to tune anything. Here are my defaults with everything vm and zfs >> related in loader.conf commented : >> >> vm.kmem_size_max: 4509713203 >> vfs.zfs.arc_max: 863907840 >> > > I was able to panic it again with "kmem_map too small" with these > settings (defaults). > > This are the bonnie++ arguments that i've used: > bonnie++ -d /tank -c 4 -r 4096 -x 9999999 -u 0:0 > > With vfs.zfs.arc_max="512M" and nothing else the machine survived more than 20 hours of bonnie++ and "iozone -a" in parallel, and is still working. Definitely the stability has improved with the new version. But maybe the algorithm for automatically setting arc_max needs a little more tweaking? (if it's not hardcoded?) - -- Regards, Nikolay Denev -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.9 (Darwin) iEYEARECAAYFAkkmczAACgkQHNAJ/fLbfrmCuACdFebPzWgigaivfOEzwXnL+2Qe 7EYAoL1KnVO7BqniD8kuZ9sPw498rILz =Yz6c -----END PGP SIGNATURE-----
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?29D6BF8D-6A05-4446-95FE-8F0AC0195C79>