Date: Mon, 3 May 2004 13:22:08 -0500 From: Dan Nelson <dnelson@allantgroup.com> To: Andrew Gallatin <gallatin@cs.duke.edu> Cc: Steve Kargl <sgk@troutmask.apl.washington.edu> Subject: Re: partial dumps (was Re: Change default dumpdir to /usr/crash?) Message-ID: <20040503182208.GB14109@dan.emsphone.com> In-Reply-To: <16534.31685.83954.524728@grasshopper.cs.duke.edu> References: <200404301403.50634.past@noc.ntua.gr> <20040430123040.GB30157@melusine.cuivre.fr.eu.org> <20040430211948.GC85783@dragon.nuxi.com> <16534.21617.310294.982202@grasshopper.cs.duke.edu> <20040503145519.GA74546@troutmask.apl.washington.edu> <p0600201cbcbc28bada78@[10.0.1.2]> <16534.31685.83954.524728@grasshopper.cs.duke.edu>
next in thread | previous in thread | raw e-mail | index | archive | help
In the last episode (May 03), Andrew Gallatin said: > Brad Knowles writes: > > At 7:55 AM -0700 2004/05/03, Steve Kargl wrote: > > > I would support the integration of the changes. With dual > > > opteron systems capacities of 16 to 24 GB of memory, partial > > > dump goes from a "nice to have feature" to a "we need this > > > feature." > > > > Hmm. What about compressed dumps? Or are we doing that already? > > savecore -z > > This would help with /var sizes, but you'd still need a dump device > as large as physmem. That's compressed vmcores :) For compressed dumps, see http://dan.allantgroup.com/FreeBSD/ . It's more than a proof-of-concept (I use it on multiple systems), but is not committable as is (see comments at the top of the diff). The ability to do kmem-only sparse dumps would reduce the necessity of true compressed dumps. -- Dan Nelson dnelson@allantgroup.com
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20040503182208.GB14109>