On 03/01/2010 08:55 AM, J. Bruce Fields wrote: > On Mon, Mar 01, 2010 at 08:48:41AM -0500, Steve Dickson wrote: >> On 03/01/2010 08:25 AM, J. Bruce Fields wrote: >>> On Mon, Mar 01, 2010 at 07:57:28AM -0500, Steve Dickson wrote: >>>> On 02/22/2010 01:22 PM, J. Bruce Fields wrote: >>>>> On Sat, Feb 20, 2010 at 03:45:10PM -0800, Trond Myklebust wrote: >>>>>> From: Trond Myklebust <Trond.Myklebust@xxxxxxxxxx> >>>>>> >>>>>> Fix a typo in commit 6d5ac3fa (nfsd: Disble NFS 4.1 functionality by default). >>>>> >>>>> Better: please just rever 6d5ac3fa and apply >>>>> http://marc.info/?l=linux-nfs&m=126540028610022&w=2 >>>> The problem I saw with the this patch was it would take >>>> a code change to re-enable the 4.1 functional, >>> >>> Enabling 4.1 will require no code changes to nfs-utils, only to the >>> kernel. (But it will require a lot of code changes to nfs-utils, as we >>> should complete 4.1 support first.) >>> >>>> verses a >>>> enabling a configuration flag. Plus, there has been a >>>> precedence set of having these types of configuration flags, >>>> note the enable_nfsv3, enable_nfsv4, so adding a enable_nfsv41 >>>> seem to me made sense... >>>> >>>>> >>>>> - distributions shouldn't be turning on 4.1 by default yet. >>>> Its not... If the --enable_nfsv41 is not set, the 4.1 functional is >>>> off. Having this type of flag enables the distros to enable the >>>> functionality on development kernels but disable it on stable >>> >>> So you want to enable it (for example) in Fedora, but disable it in >>> RHEL? >> No, its only enabled in Rawhide which is the untested/developmental >> version of the next fedora release (in this case F14). When F14 is ready, >> the determination will be made (with solicited input) if the 4.1 code >> should continue to be enabled or disabled. > > That decision should depend on whether the kernel 4.1 support is > complete or not. If it is, then the kernel will default 4.1 on, and > there will be no need for an nfs-utils change. > >> This is why I like the configuration switch. There is no code >> changes, code is just being disabled. Which, in theory, should have >> no ramification at all.. >> >>> >>> I don't think that's a good idea. If servers running current Fedora >>> kernels are still around when the client starts trying 4.1 first, then >>> we'll run into trouble. >> Since people actually have to specify the -o minorversion=1 mount >> option (which is not documented, btw), I'm going to assume those people >> know what they are doing and probably will know how to fix any problems >> that may arise... > > No, that's not enough. They may be accessing the server through other > clients. Understood... just curious, would you happen to know of any clients that use 4.1 as the default? Was there any at this year's Connectathon? > > We need users to make a conscious decision to turn on the server-side. > It's not enough to do it only on the client side. Right. At this point people will either have to recompile nfs-utils using the --enable-nfsv41 flag or edit their initscript and do the echo trick... So isn't having to do one of the above be construed as conscious decision? steved. -- To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
