Date: Fri, 16 Jan 2009 10:32:29 +0300 From: Lev Serebryakov <lev@serebryakov.spb.ru> To: "Rick C. Petty" <rick-freebsd2008@kiwi-computer.com> Cc: Brian McCann <bjmccann@gmail.com>, freebsd-geom@freebsd.org Subject: Re[2]: gvinum & gjournal Message-ID: <529173009.20090116103229@serebryakov.spb.ru> In-Reply-To: <20090115172036.GA54383@keira.kiwi-computer.com> References: <2b5f066d0901141323j7c9a194eo4606d9769279037e@mail.gmail.com> <20090115025645.21ad2185.ota@j.email.ne.jp> <2b5f066d0901150410s7dc4e97v741d5edd2a4983a9@mail.gmail.com> <20090115172036.GA54383@keira.kiwi-computer.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Hello, Rick. You wrote 15 =FF=ED=E2=E0=F0=FF 2009 =E3., 20:20:36: > You don't *have* to fsck with UFS2 either, if you're using soft updates. > The only thing fsck does is free up space and inodes that are marked as > used but are really not used. Since it can be done successfully in the > background, I don't see much of a problem (yes it will take hours, so > schedule the checks at times when you have the least I/O traffic). background fsck on 2Tb (RAID-5 on 5x500Gb drives) in same time as raid5 rebuilding (due to same reasons as fsck: dirty reboot) is pain in ass, really. It finishes never. I was need to stop RAID5 rebuilding, umount filesystem, run fsck by hands (with lots of questions about strange softupdate inconsistences) TWICE and rebuild RAID5 after that. Only thing I lost is 6 or 7 files which were "in flight" at time of crash. So, yes, it is safe for information (nobody expect to have files in flight to be safe in case of crash, of course) but it does not work automagically after reboot... I'm thinking about gjournal, but it KILLS performance in case of fast RAID array without additional super-fast (SSD?) place for journal :( --=20 // Black Lion AKA Lev Serebryakov <lev@serebryakov.spb.ru>
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?529173009.20090116103229>