Date: Sat, 31 May 2008 11:47:33 -0700 From: Jeremy Chadwick <koitsu@FreeBSD.org> To: John Baldwin <jhb@freebsd.org> Cc: freebsd-hackers@freebsd.org, Tz-Huan Huang <tzhuan@csie.org> Subject: Re: Is there any way to increase the KVM? Message-ID: <20080531184733.GA8470@eos.sc1.parodius.com> In-Reply-To: <200805311249.53986.jhb@freebsd.org> References: <6a7033710805302252v43a7b240x66ca3f5e3dd5fda4@mail.gmail.com> <200805311249.53986.jhb@freebsd.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Sat, May 31, 2008 at 12:49:53PM -0400, John Baldwin wrote: > On Saturday 31 May 2008 01:52:56 am Tz-Huan Huang wrote: > > Hi, > > > > Our nfs server is running 7-stable/amd64 with 8G ram, the size of zfs > > pool is 12T. We have set vm.kmem_size and vm.kmem_size_max to > > 1.5G, but the kernel still panics by "kmem_map too small" often. > > According to [1], the limitation is not only by the loader (is it fixed > > now?) but also by the default layout of KVM. [2] points a way to increase > > the KVM, but we get the similar linking error. > > > > Is there any standard way to modify the layout of KVM? For example, we > > may want to set KVM to 6G and leave the 2G for user space usage. > > On i386 you only have 4GB of virtual address space period. For amd64 you can > jack up KVM just fine AFAIK. The mcmodel=kernel stuff should only affect > global variables (so .data and .bss) and not malloc'd stuff. Have you tried > increasing the KVM size and seeing what happens? On amd64 there's still a 2GB limit, re: vm.kmem_size. Increasing it to or past 2048M results in the message "kmem_suballoc(): bad status return of 3". http://lists.freebsd.org/pipermail/freebsd-stable/2008-May/042764.html -- | Jeremy Chadwick jdc at parodius.com | | Parodius Networking http://www.parodius.com/ | | UNIX Systems Administrator Mountain View, CA, USA | | Making life hard for others since 1977. PGP: 4BD6C0CB |
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20080531184733.GA8470>