Date: Wed, 2 Dec 2015 14:00:19 +0000 From: krad <kraduk@gmail.com> To: Tom Evans <tevans.uk@googlemail.com> Cc: Zeus Panchenko <zeus@ibs.dn.ua>, FreeBSD Filesystems <freebsd-fs@freebsd.org> Subject: Re: advice needed: zpool of 10 x (raidz2 on (4+2) x 2T HDD) Message-ID: <CALfReycBOY3i3HRLqNVTChC=fL-=KGJT4eXBhN_=R8uGXQtSJA@mail.gmail.com> In-Reply-To: <CAFHbX1%2B9JiGeukjePqzy1vTS1-ON1vY27v2iB5daX1XGjpYNDA@mail.gmail.com> References: <20151202133428.35820@smtp.new-ukraine.org> <CAFHbX1%2B9JiGeukjePqzy1vTS1-ON1vY27v2iB5daX1XGjpYNDA@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
If this is the case, put more ram in and a good amount of SSD for l2arc. SSD is fairly cheap, so put in a TB of the best you can afford. RAM is always best though if you can afford it and fit it onto the board. It's probably worth going to 10.2 as well, and maybe even stable as there have been a lot of zfs related patches since 10.1. On 2 December 2015 at 13:13, Tom Evans via freebsd-fs < freebsd-fs@freebsd.org> wrote: > On Wed, Dec 2, 2015 at 11:34 AM, Zeus Panchenko <zeus@ibs.dn.ua> wrote: > > -----BEGIN PGP SIGNED MESSAGE----- > > Hash: SHA1 > > > > greetings, > > > > we deployed storage, and as it was filling until now, I see I need > > an advice regarding the configuration and optimization/s ... > > > > the main cause I decided to ask for an advice is this: > > > > once per month (or even more frequently, depends on the load I > > suggest) host hangs and only power reset helps, nothing helpful in log > > files though ... just the fact of restart logged and usual ctld activity > > <snip> > > CPU: Intel(R) Xeon(R) CPU E5-2630L (2 package(s) x 6 core(s) x 2 SMT > threads) > > RAM: 128Gb > > STOR: 3 x LSI SAS2308 PCI-Express Fusion-MPT SAS-2 (jbod) > > 60 x HDD 2T (ATA WDC WD20EFRX-68A 0A80, Fixed Direct Access > SCSI-6 device 600.000MB/s) > > > > OS: FreeBSD 10.1-RELEASE #0 r274401 amd64 > > > > to avoid OS memory shortage sysctl vfs.zfs.arc_max is set to 120275861504 > > > > to clients, storage is provided via iSCSI by ctld (each target is file > backed) > > > > zpool created of 10 x raidz2, each raidz2 consists of 6 geli devices and > > now looks so (yes, deduplication is on): > > > >> zpool list storage > > NAME SIZE ALLOC FREE FRAG EXPANDSZ CAP DEDUP > HEALTH ALTROOT > > storage 109T 33.5T 75.2T - - 30% 1.57x > ONLINE - > > You will need to examine zdb output to correctly determine the size of > your dedupe table. Assuming an average block size of 64kb, your DDT > will be approximately 167GB, ie well outside your RAM (my maths may be > off of course - ((33.5*(2**40)/(64*1024))*320)/(2**30)). > > This article explains in detail: > > http://constantin.glez.de/blog/2011/07/zfs-dedupe-or-not-dedupe > > Cheers > > Tom > _______________________________________________ > freebsd-fs@freebsd.org mailing list > https://lists.freebsd.org/mailman/listinfo/freebsd-fs > To unsubscribe, send any mail to "freebsd-fs-unsubscribe@freebsd.org" >
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CALfReycBOY3i3HRLqNVTChC=fL-=KGJT4eXBhN_=R8uGXQtSJA>