Re: Fix to diskless(8) manpage: add amd64 and i386 to the list of clients that needs rpc.bootparamd(8)
On 7/14/13, Rafael Neves rafaelne...@gmail.com wrote: On 7/14/13, Theo de Raadt dera...@cvs.openbsd.org wrote: Date: Sun, 14 Jul 2013 18:51:13 +0200 From: Sebastian Benoit be...@openbsd.org Mark Kettenis(mark.kette...@xs4all.nl) on 2013.07.14 17:06:24 +0200: ... Index: diskless.8 === RCS file: /home/cvs/src/share/man/man8/diskless.8,v retrieving revision 1.61 diff -u -p -r1.61 diskless.8 --- diskless.8 15 Apr 2013 06:29:57 - 1.61 +++ diskless.8 14 Jul 2013 17:03:47 - @@ -364,8 +364,9 @@ For all clients: .Xr mountd 8 , .Xr nfsd 8 , .Xr portmap 8 , +.Xr rarpd 8 , and -.Xr rarpd 8 . +.Xr rpc.bootparamd 8 . .Pp For alpha, amd64, hppa, hppa64, i386, sgi, mvme68k, mvme88k, sparc, sparc64, and vax clients: @@ -376,9 +377,6 @@ For HP 300 and older HPPA clients: .Pp For newer alpha, amd64, hppa, hppa64, i386, and sgi clients: .Xr dhcpd 8 -.Pp -For hp300, older hppa, mvme68k, mvme88k, sparc, sparc64, and vax clients: -.Xr rpc.bootparamd 8 .Pp For older alpha and vax clients: .Xr mopd 8 Makes sense, is shorter and clearer. Feel free to adopt any version that you and devs think is more suitable. I concur. diskless(8) is not really a place to explain every detail about why, and quite frankly when a boot block goes and uses dhcp information knowing full well rpc.bootparams will be needed later, you can guess what I consider the problem.. Hi Mark, Just remembering about this diff. Regards Rafael Neves
Fix to diskless(8) manpage: add amd64 and i386 to the list of clients that needs rpc.bootparamd(8)
Hi, Amd64 and i386 diskless(8) setups need rpc.bootparamd(8) to be enabled to read /etc/bootparams. The patch below adds both archs to the list of the ones that needs rpc.bootparamd(8). Index: diskless.8 === RCS file: /cvs/src/share/man/man8/diskless.8,v retrieving revision 1.61 diff -u -p -r1.61 diskless.8 --- diskless.8 15 Apr 2013 06:29:57 - 1.61 +++ diskless.8 14 Jul 2013 08:23:35 - @@ -377,7 +377,7 @@ For HP 300 and older HPPA clients: For newer alpha, amd64, hppa, hppa64, i386, and sgi clients: .Xr dhcpd 8 .Pp -For hp300, older hppa, mvme68k, mvme88k, sparc, sparc64, and vax clients: +For amd64, hp300, older hppa, i386, mvme68k, mvme88k, sparc, sparc64, and vax clients: .Xr rpc.bootparamd 8 .Pp For older alpha and vax clients:
Re: Fix to diskless(8) manpage: add amd64 and i386 to the list of clients that needs rpc.bootparamd(8)
On Sun, Jul 14, 2013 at 05:59:53AM -0300, Rafael Neves wrote: Hi, Amd64 and i386 diskless(8) setups need rpc.bootparamd(8) to be enabled to read /etc/bootparams. The patch below adds both archs to the list of the ones that needs rpc.bootparamd(8). they shouldn;t - this stuff is all covered by step 4 in diskless(8). amd64 and i386 use dhcpd, and shouldn;t need rpc.bootparamd. it's a long time since i tested it, but i'm pretty sure it works without. jmc Index: diskless.8 === RCS file: /cvs/src/share/man/man8/diskless.8,v retrieving revision 1.61 diff -u -p -r1.61 diskless.8 --- diskless.815 Apr 2013 06:29:57 - 1.61 +++ diskless.814 Jul 2013 08:23:35 - @@ -377,7 +377,7 @@ For HP 300 and older HPPA clients: For newer alpha, amd64, hppa, hppa64, i386, and sgi clients: .Xr dhcpd 8 .Pp -For hp300, older hppa, mvme68k, mvme88k, sparc, sparc64, and vax clients: +For amd64, hp300, older hppa, i386, mvme68k, mvme88k, sparc, sparc64, and vax clients: .Xr rpc.bootparamd 8 .Pp For older alpha and vax clients:
Re: Fix to diskless(8) manpage: add amd64 and i386 to the list of clients that needs rpc.bootparamd(8)
Mark Kettenis(mark.kette...@xs4all.nl) on 2013.07.14 17:06:24 +0200: Date: Sun, 14 Jul 2013 14:09:26 +0200 From: Henning Brauer lists-openbsdt...@bsws.de * Rafael Neves rafaelne...@gmail.com [2013-07-14 11:01]: Amd64 and i386 diskless(8) setups need rpc.bootparamd(8) no, they don't. True diskless(4) operation (with root on nfs) needs rpc.bootparamd(8) an *all* architectures. Merely booting a bsd.rd kernel only needs rpc.bootparamd(8) on the architectures mentioned under 11 in the EXAMPLES section of the diskless(4) page. yes, and having run though diskless 4 months ago and discovering that it was missing, i tought about a similar manpage diff. I mean, the page starts with When booting a system over the network, there are three phases of interaction between client and server: 1. The PROM (or stage-1 bootstrap) loads a boot program. 2. The boot program loads a kernel. 3. The kernel does NFS mounts for root and swap. and i would expect thats what i get when doing what it describes. Maybe just add this? --- diskless.8~ Sun Jul 14 18:44:42 2013 +++ diskless.8 Sun Jul 14 18:44:42 2013 @@ -382,6 +382,10 @@ .Pp For older alpha and vax clients: .Xr mopd 8 +.Pp +All architectures need +.Xr rpc.bootparamd 8 +to retrieve the location of the root and swap device. .It Net boot the client. .El
Re: Fix to diskless(8) manpage: add amd64 and i386 to the list of clients that needs rpc.bootparamd(8)
Date: Sun, 14 Jul 2013 18:51:13 +0200 From: Sebastian Benoit be...@openbsd.org Mark Kettenis(mark.kette...@xs4all.nl) on 2013.07.14 17:06:24 +0200: Date: Sun, 14 Jul 2013 14:09:26 +0200 From: Henning Brauer lists-openbsdt...@bsws.de * Rafael Neves rafaelne...@gmail.com [2013-07-14 11:01]: Amd64 and i386 diskless(8) setups need rpc.bootparamd(8) no, they don't. True diskless(4) operation (with root on nfs) needs rpc.bootparamd(8) an *all* architectures. Merely booting a bsd.rd kernel only needs rpc.bootparamd(8) on the architectures mentioned under 11 in the EXAMPLES section of the diskless(4) page. yes, and having run though diskless 4 months ago and discovering that it was missing, i tought about a similar manpage diff. I mean, the page starts with When booting a system over the network, there are three phases of interaction between client and server: 1. The PROM (or stage-1 bootstrap) loads a boot program. 2. The boot program loads a kernel. 3. The kernel does NFS mounts for root and swap. and i would expect thats what i get when doing what it describes. Maybe just add this? I don't think that's particularly helpful. We should just add rpc.bootparamd(8) to the For all clients list. The pxebooting crowd that only wants to boot an installation kernel should be pointed at pxeboot(8) instead of diskless(8). Or they'll just do a little bit of extra work ;). Index: diskless.8 === RCS file: /home/cvs/src/share/man/man8/diskless.8,v retrieving revision 1.61 diff -u -p -r1.61 diskless.8 --- diskless.8 15 Apr 2013 06:29:57 - 1.61 +++ diskless.8 14 Jul 2013 17:03:47 - @@ -364,8 +364,9 @@ For all clients: .Xr mountd 8 , .Xr nfsd 8 , .Xr portmap 8 , +.Xr rarpd 8 , and -.Xr rarpd 8 . +.Xr rpc.bootparamd 8 . .Pp For alpha, amd64, hppa, hppa64, i386, sgi, mvme68k, mvme88k, sparc, sparc64, and vax clients: @@ -376,9 +377,6 @@ For HP 300 and older HPPA clients: .Pp For newer alpha, amd64, hppa, hppa64, i386, and sgi clients: .Xr dhcpd 8 -.Pp -For hp300, older hppa, mvme68k, mvme88k, sparc, sparc64, and vax clients: -.Xr rpc.bootparamd 8 .Pp For older alpha and vax clients: .Xr mopd 8
Re: Fix to diskless(8) manpage: add amd64 and i386 to the list of clients that needs rpc.bootparamd(8)
Date: Sun, 14 Jul 2013 18:51:13 +0200 From: Sebastian Benoit be...@openbsd.org Mark Kettenis(mark.kette...@xs4all.nl) on 2013.07.14 17:06:24 +0200: Date: Sun, 14 Jul 2013 14:09:26 +0200 From: Henning Brauer lists-openbsdt...@bsws.de * Rafael Neves rafaelne...@gmail.com [2013-07-14 11:01]: Amd64 and i386 diskless(8) setups need rpc.bootparamd(8) no, they don't. True diskless(4) operation (with root on nfs) needs rpc.bootparamd(8) an *all* architectures. Merely booting a bsd.rd kernel only needs rpc.bootparamd(8) on the architectures mentioned under 11 in the EXAMPLES section of the diskless(4) page. yes, and having run though diskless 4 months ago and discovering that it was missing, i tought about a similar manpage diff. I mean, the page starts with When booting a system over the network, there are three phases of interaction between client and server: 1. The PROM (or stage-1 bootstrap) loads a boot program. 2. The boot program loads a kernel. 3. The kernel does NFS mounts for root and swap. and i would expect thats what i get when doing what it describes. Maybe just add this? I don't think that's particularly helpful. We should just add rpc.bootparamd(8) to the For all clients list. The pxebooting crowd that only wants to boot an installation kernel should be pointed at pxeboot(8) instead of diskless(8). Or they'll just do a little bit of extra work ;). Index: diskless.8 === RCS file: /home/cvs/src/share/man/man8/diskless.8,v retrieving revision 1.61 diff -u -p -r1.61 diskless.8 --- diskless.815 Apr 2013 06:29:57 - 1.61 +++ diskless.814 Jul 2013 17:03:47 - @@ -364,8 +364,9 @@ For all clients: .Xr mountd 8 , .Xr nfsd 8 , .Xr portmap 8 , +.Xr rarpd 8 , and -.Xr rarpd 8 . +.Xr rpc.bootparamd 8 . .Pp For alpha, amd64, hppa, hppa64, i386, sgi, mvme68k, mvme88k, sparc, sparc64, and vax clients: @@ -376,9 +377,6 @@ For HP 300 and older HPPA clients: .Pp For newer alpha, amd64, hppa, hppa64, i386, and sgi clients: .Xr dhcpd 8 -.Pp -For hp300, older hppa, mvme68k, mvme88k, sparc, sparc64, and vax clients: -.Xr rpc.bootparamd 8 .Pp For older alpha and vax clients: .Xr mopd 8 I concur. diskless(8) is not really a place to explain every detail about why, and quite frankly when a boot block goes and uses dhcp information knowing full well rpc.bootparams will be needed later, you can guess what I consider the problem..
Re: Fix to diskless(8) manpage: add amd64 and i386 to the list of clients that needs rpc.bootparamd(8)
On 7/14/13, Theo de Raadt dera...@cvs.openbsd.org wrote: Date: Sun, 14 Jul 2013 18:51:13 +0200 From: Sebastian Benoit be...@openbsd.org Mark Kettenis(mark.kette...@xs4all.nl) on 2013.07.14 17:06:24 +0200: Date: Sun, 14 Jul 2013 14:09:26 +0200 From: Henning Brauer lists-openbsdt...@bsws.de * Rafael Neves rafaelne...@gmail.com [2013-07-14 11:01]: Amd64 and i386 diskless(8) setups need rpc.bootparamd(8) no, they don't. True diskless(4) operation (with root on nfs) needs rpc.bootparamd(8) an *all* architectures. Merely booting a bsd.rd kernel only needs rpc.bootparamd(8) on the architectures mentioned under 11 in the EXAMPLES section of the diskless(4) page. Exactly, the issue show up only when booting bsd kernel with root on nfs. Actually, bsd.rd needs only dhcpd(8) and tftpd(8). Thanks for the accurate description. yes, and having run though diskless 4 months ago and discovering that it was missing, i tought about a similar manpage diff. I mean, the page starts with When booting a system over the network, there are three phases of interaction between client and server: 1. The PROM (or stage-1 bootstrap) loads a boot program. 2. The boot program loads a kernel. 3. The kernel does NFS mounts for root and swap. and i would expect thats what i get when doing what it describes. Maybe just add this? I don't think that's particularly helpful. We should just add rpc.bootparamd(8) to the For all clients list. The pxebooting crowd that only wants to boot an installation kernel should be pointed at pxeboot(8) instead of diskless(8). Or they'll just do a little bit of extra work ;). Index: diskless.8 === RCS file: /home/cvs/src/share/man/man8/diskless.8,v retrieving revision 1.61 diff -u -p -r1.61 diskless.8 --- diskless.8 15 Apr 2013 06:29:57 - 1.61 +++ diskless.8 14 Jul 2013 17:03:47 - @@ -364,8 +364,9 @@ For all clients: .Xr mountd 8 , .Xr nfsd 8 , .Xr portmap 8 , +.Xr rarpd 8 , and -.Xr rarpd 8 . +.Xr rpc.bootparamd 8 . .Pp For alpha, amd64, hppa, hppa64, i386, sgi, mvme68k, mvme88k, sparc, sparc64, and vax clients: @@ -376,9 +377,6 @@ For HP 300 and older HPPA clients: .Pp For newer alpha, amd64, hppa, hppa64, i386, and sgi clients: .Xr dhcpd 8 -.Pp -For hp300, older hppa, mvme68k, mvme88k, sparc, sparc64, and vax clients: -.Xr rpc.bootparamd 8 .Pp For older alpha and vax clients: .Xr mopd 8 Makes sense, is shorter and clearer. Feel free to adopt any version that you and devs think is more suitable. I concur. diskless(8) is not really a place to explain every detail about why, and quite frankly when a boot block goes and uses dhcp information knowing full well rpc.bootparams will be needed later, you can guess what I consider the problem..