Feature changed by: Antoine Ginies (aginies) Feature #315589, revision 16 Title: [BETA 1] Deprecate bind-mount configuration with NFSv4 export Requested by: Neil Brown (neilbrown) Partner organization: openSUSE.org Description: When NFSv4 was introduced it was not possible to export filesystems to NFSv4 the same way as for NFSv3. The "bind=" syntax was added to /etc/exports to help bridge the gaps. With current kernel and nfs-utils this is no longer an issue. By default any filesystem that is exported is exported equally to NFSv2, v3, and v4. So we should deprecate the "bind=" syntax and eventually remove it. This means: 1) yasts nfs-server module should remove knowledge of "Bindmount targets". The "bind=..." option should just be treated like any other option. 2) yasts nfs-server module should not encourage the "fsid=0" export option. Like "bind=" it must still be permitted, but it should be considered and exception rather than normal. 3) Yasts nfs-server help screens should remove mention of "bind=" and "fsid=0". 4) Any documentation that mentioned nfs export, such as http://doc.opensuse.org/documentation/html/openSUSE/opensuse-reference/cha.n... should be changed to remove mention of bind= and fsid=, and should not make any distinciton between NFSv4 and NFSv3. Documentation Impact: - RN Admin Guide, net_nfs.xml Test Case: 1) start yast2-nfs-server 2) try to create a NFS4 enabled export 3) fsid=0 and / or bind= options should NOT be present in provided default configuration 4) try to mount the export remotely Business case (Partner benefit): openSUSE.org: Simplifies administration, reduces possible confusion, remove unnecessary difference between SUSE and other distros. Discussion: #3: Lukas Ocilka (locilka) (2013-08-23 12:39:57) Michal, this rather seems to be a low-hanging fruit. Please, proceed. #4: Michal Filka (mfilka) (2014-01-24 11:19:02) no done yet. Postponed due to work on refactoring yast2-network and new installer #5: Karl Eichwalder (keichwa) (2014-04-28 17:25:07) What about "Coexisting v3 and v4 Exports"? We have such a statement in our docs: NFSv3 and NFSv4 exports can coexist on a server. After enabling the support for NFSv4 in the initial configuration dialog, those exports for which 'fsid=0' and 'bind=/target/path' are not included in the option list are considered v3 exports. ============================================================ I'm also wondering what actually happens if you "Enable NFSv4" in yast, or not. If you "Enable NFSv4", will yast then support both, v3 and v4? #6: Neil Brown (neilbrown) (2014-04-29 17:15:51) (reply to #5) We should probably revise those docs. There is no longer an important difference between NFSv3 and NFSv4 exports. If you don't have "fsid=0" or and "bind=...." in /etc/exports then all filesystems mentioned are exported equally via NFSv3 and NFSv4. If "fsid=0" is present then the old approach still works for backward compatibility, but we really want to avoud using that whenever possible. #7: Karl Eichwalder (keichwa) (2014-04-29 15:21:03) (reply to #6) Thanks, I changed the text in the docs accordingly. And I added a minimal release notes snippet. Release Notes: Exporting NFSv4 Shares Challenge: On SUSE Linux Enterprise 11, the bind mount in /etc/exports was mandatory. It is still supported, but now deprecated. Solution: Configuring directories for export with NFSv4 is now the same as with NFSv3. -- openSUSE Feature: https://features.opensuse.org/315589