Date: Thu, 03 Apr 2008 23:15:05 +0200 From: Volker <volker@vwsoft.com> To: Norberto Meijome <freebsd@meijome.net> Cc: freebsd-stable@freebsd.org Subject: Re: Q&A on textdumps (fwd) Message-ID: <47F548D9.8060905@vwsoft.com> In-Reply-To: <20080404012211.2bff6dd2@meijome.net> References: <20080401125534.D94491@fledge.watson.org> <20080402131511.2940ef2e@meijome.net> <47F3E3D8.4080308@vwsoft.com> <20080404012211.2bff6dd2@meijome.net>
next in thread | previous in thread | raw e-mail | index | archive | help
On 04/03/08 16:22, Norberto Meijome wrote: > On Wed, 02 Apr 2008 21:51:52 +0200 > Volker <volker@vwsoft.com> wrote: > >> while the system is experiencing a panic, it does not have any knowledge >> about filesystems and also does not know about the GELI swap space anymore. >> >> In this situation the geli encrypted swap will be overwritten by a dump >> (either minidump or the classical one). When the system boots up again, >> it will check $dumpdev for a dump and save it to $savecore before geli >> swap is brought up again. >> >> Or in short: geli backed swap should not do you any harm. > > Hi Volker, > That is what I had thought. But I clearly remember, back in 6.x the dump would be written to swap, then on restart, geli would kick in before savedump , therefore obliterating anything saved from the previous session. > > I will need to retest this...luckily 7-STABLE has been rock solid and haven't experienced many crashes. Hi Norberto, you may check the start order by using: `rcorder /etc/rc.d/* /usr/local/etc/rc.d/*' On my 7-stable system, the interesting ones are in this order: dumpon geli encswap swap1 savecore ...ouch! ;) This does not look like the order it should be but I think (fortunately) in most situations this should be ok as nothing gets written to swap space before daemon processes are brought up. On the other side, savecore needs mounted filesystems to save the core to /var/crash. Volker
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?47F548D9.8060905>