Date: Wed, 31 Dec 2003 15:44:32 -0800 From: Tim Kientzle <kientzle@acm.org> To: Doug Barton <DougB@freebsd.org> Cc: David Gilbert <dgilbert@dclg.ca> Subject: Re: 5.1-R-p11 unable to "sync disks" when shutting down Message-ID: <3FF35F60.1070803@acm.org> In-Reply-To: <20031231152453.G1441@bo.vpnaa.bet> References: <20031229164130.GA797@pp.htv.fi> <16370.20382.81302.449649@canoe.dclg.ca> <16371.20373.212950.533985@canoe.dclg.ca> <20031231152453.G1441@bo.vpnaa.bet>
next in thread | previous in thread | raw e-mail | index | archive | help
Doug Barton wrote: > On Wed, 31 Dec 2003, David Gilbert wrote: >>I'd like to see: >> >>crash -> boot -> defer for bgfsck (but mark disk as having tried) >> -> multiuser -> bgfsck -> crash -> boot -> fsck in foreground >> -> multiuser >> >>... but I believe this would require fsck support unless the startup >>scripts were writing to the filesystem to indicate no bgfsck. > > > What you propose would also be really complicated for other reasons > related to the way rcNG works. I agree that it would be tricky to handle within the startup scripts. But it doesn't seem like it should be hard to handle within fsck itself. The first time fsck gets invoked by rcNG, it is given the -F flag which, according to fsck(8) does a foreground check on "all the file systems that cannot do background checking." I think that "system rebooted during a bg fsck" could qualify that FS as being unable to do background checking. What's needed is some way for fsck to mark a filesystem as having a bgfsck in progress. Then fsck -F can check for that marker. Tim
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?3FF35F60.1070803>