Date: Mon, 27 Jan 2014 16:52:17 +0000 From: krad <kraduk@gmail.com> To: Kaya Saman <kayasaman@gmail.com> Cc: freebsd-questions <freebsd-questions@freebsd.org>, =?ISO-8859-1?Q?Trond_Endrest=F8l?= <Trond.Endrestol@fagskolen.gjovik.no> Subject: Re: ZFS confusion Message-ID: <CALfReyf_V%2B9iWrPTceHya9GnW3c6_aNA%2BEt8ZzJQypNsHrriNQ@mail.gmail.com> In-Reply-To: <52E6657E.1050103@gmail.com> References: <52E40C82.7050302@gmail.com> <alpine.BSF.2.00.1401270944100.4811@mail.fig.ol.no> <52E62DFF.3010600@gmail.com> <alpine.BSF.2.00.1401271149160.4811@mail.fig.ol.no> <52E6463C.6090805@gmail.com> <alpine.BSF.2.00.1401271309090.4811@mail.fig.ol.no> <52E6537F.8020907@gmail.com> <CALfReyeO0sOAHLp2y_FpxT-yAUoqSdrtdbjQG_EEOzNL33q1XQ@mail.gmail.com> <52E6657E.1050103@gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Look into under provisioning the SSD drives as well, this can preserve write performance in the long term and decrease write wear. Looking at the number of drives, and general spec of what you are putting together, I would try to stretch to 256 GB ssd but only provision them to use say 128-160 GB of the capacity. I'm not 100% sure this is still all necessary now as TRIM support is much better now under zfs but here is how i did my ssd drives under linux. You may well be able to do it under freebsd but I havent figured out how. root@ubuntu-10-10:~# hdparm -N /dev/sdb /dev/sdb: max sectors = 312581808/312581808, HPA is disabled root@ubuntu-10-10:~# hdparm -Np281323627 /dev/sdb /dev/sdb: setting max visible sectors to 281323627 (permanent) Use of -Nnnnnn is VERY DANGEROUS. You have requested reducing the apparent size of the drive. This is a BAD idea, and can easily destroy all of the drive's contents. Please supply the --yes-i-know-what-i-am-doing flag if you really want this. Program aborted. root@ubuntu-10-10:~# hdparm -Np281323627 --yes-i-know-what-i-am-doing /dev/sdb /dev/sdb: setting max visible sectors to 281323627 (permanent) max sectors = 281323627/312581808, HPA is enabled root@ubuntu-10-10:~# On 27 January 2014 13:56, Kaya Saman <kayasaman@gmail.com> wrote: > Many thanks for the explanation :-) > > > On 01/27/2014 01:13 PM, krad wrote: > >> Neither of these setups is ideal, The best practice for your vdev is to >> use 2^n + your parity drives >> This means in your case with raidz3 you would do something >> >> 2 + 3 >> 4 + 3 >> 8 + 3 >> >> the 1st two are far from ideal as the ratios are low 8 + 3, so 11 drives >> per raidz3 vdev would be optimal. This would fit nicely with your 26 drive >> enclosure as you would use 2x11 drive raidz3 vdevs, 2 hot spares, and two >> devices left for l2arc/zil. Probably best chop up the ssds, mirror the zil >> and stripe the l2arc, assuming you dont want to do down the route using >> generic SSD's rather than write/read optimized ones >> > > Yep was going to use your suggestion for l2arc/zil on 2x 128GB Corsair > Force Series GS, 2.5" which have quite good w/r speeds - also I use these > on other servers which tend to be quite good and reliable. > > I think the way to create a mirrored zil and stiped l2arc would be to use > GPT to partition the drives, then use the zfs features across the > partitions. > > > Hmm.... so it also looks like I'm gona have to wait a while for some more > drives then in order to create an 11 disk raidz3 pool. > > > But at least things will be done properly and in a good manner rather then > going down a patch with "no return". > > > >> >> Regards, > > > Kaya >
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CALfReyf_V%2B9iWrPTceHya9GnW3c6_aNA%2BEt8ZzJQypNsHrriNQ>