Date: Thu, 21 Jun 2012 16:52:25 +0200 (CEST) From: Wojciech Puchar <wojtek@wojtek.tensor.gdynia.pl> To: Eduardo Morras <nec556@retena.com> Cc: freebsd-questions@freebsd.org Subject: Re: Is ZFS production ready? Message-ID: <alpine.BSF.2.00.1206211647100.3170@wojtek.tensor.gdynia.pl> In-Reply-To: <4FA8826F00BC43C3@> (added by postmaster@resmaa12.ono.com) References: <4FA8826F00BC43C3@> (added by postmaster@resmaa12.ono.com)
next in thread | previous in thread | raw e-mail | index | archive | help
> > With UFS2 you can use file systems up to 2^73 (8 ZB). The problem is not UFS, > but the old tools used to format the disk like fdisk and bsdlabel. For big > file systems you must use gpart. true. or not using anything at all (and put filesystem directly on whole device/mirror). > The problem with file system recovery times when "the worst thing > happens"(tm) is soluted/mitigated with su+j on FreeBSD9. True but i don't believe completely in SU+J. i use it - eg on my private backup disk. but do full fsck sometimes. and usually few, but nonzero amount of errors are corrected. but with just SU it is easy to solve. Disable fsck on boot at all. softupdates allow that risk without problems. then do fsck at time when full or partial system outage can be tolerated - after work hours. This is my solution used everywhere. of course fsck on 100TB filesystem will be too slow. But it is implementation problem, and could be improved. but i would not recommend making single virtual device (gmirror/gstripe or dedicated hardware matrix controller) from too many disks because of the risk.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?alpine.BSF.2.00.1206211647100.3170>