https://jira.whamcloud.com This file is an XML representation of an issue en-us 9.4.14 940014 05-12-2023 https://jira.whamcloud.com/browse/LU-28 Lustre <p>In kicking around some ideas for the use of Lustre as a replacement for NFS user directories (home and application areas), the thought of being able to export/mount a subdirectory of a Lustre LOV came up.</p> LU-28

Ability for client to mount subdirectories of a Lustre filesystem

Bug Minor Resolved Fixed Wang Shilong Lai Siyao patch Mon, 20 Dec 2010 17:50:17 +0000 Fri, 23 Sep 2016 13:05:16 +0000 Mon, 18 Apr 2016 12:55:49 +0000 Lustre 2.0.0 Lustre 2.9.0 Fri, 28 Jan 2011 00:00:00 +0000 0 19 <p>Basically there are two ways to implement this:<br/> 1. client lookup the filesystem root, and then walk the filesystem (based on the<br/> subdirectory path) and instantiate the subdirectory as the mountpoint root.<br/> This can be implemented entirely on the client and does not require any protocol<br/> changes.<br/> 2. change the MDS_GETSTATUS RPC to pass the subdirectory pathname from the client<br/> to the MDS, and have it do the pathname traversal locally and return the "root"<br/> FID for the subdirectory. This needs protocol changes, though simple ones that<br/> can be implemented in a backward-complatible manner.</p> <p>The advantage of the first way is:</p> <ul> <li>no protocol changes.</li> <li>client changes only, no compatibility issue.</li> <li>pathname transversal is more convenient and natural in client, especially for CMD<br/> which may involve several MDS.</li> </ul> <p>Due to the above advantages this feature will be implemented in the first way.</p> <p>However there are some limitations:</p> <ul> <li>&lt;mountpoint&gt;/.lustre doesn't exist for subdirectory mount. This '.lustre' directory<br/> is maintained under root directory on MDS for whole filesystem. Though we may create<br/> a link under subdirectory pointing to it, it may leak fids outside of its namespace.<br/> Currently the only affected utility is lustre_rsync, but user can do a normal mount<br/> and apply there (actually lustre_rsync should work on a normal mount, otherwise it<br/> may not be able to access files out of its namespace).</li> <li>symlink in the subdirectory will not be supported due to the extra complexity.</li> </ul> <p>The following part of code will be changed:</p> <ul> <li>llite mount transverse subdirectory path.</li> <li>`lfs fid2path` support subdirectory mount.</li> </ul> <p>code ready and is being reviewed by Niu and Bobi.</p> <p>Has there been any progress on this issue? I haven't seen anything, but I thought I'd check as people have been asking about it.</p> <p>The original patch is at <a href="http://review.whamcloud.com/#change,165" class="external-link" target="_blank" rel="nofollow noopener">http://review.whamcloud.com/#change,165</a>, but it's not up to date. Since it's not urgent, I don't have plan to commit it soon. If someone would like to take over this work, that would be great.</p> <p>I rebased it on master, it's here: <a href="http://review.whamcloud.com/#change,5007" class="external-link" target="_blank" rel="nofollow noopener">http://review.whamcloud.com/#change,5007</a></p> <p>Thanks for the review. I'm still new to Gerrit, is there someone I should add for +2?</p> <p>Kit</p> <p>No further action is needed - the patch is already in the queue for landing consideration</p> <p>Peter</p> <p>Great, thanks</p> <p>What would also be very useful here for future expansion is to enhance the MDS_GETSTATUS RPC to accept a fileset name, currently always "ROOT", so that this handling can be done by the MDS in the future and the "traverse directory on client" hack can be avoided. That would allow completely independent filesets to be handled by different subtrees on the same MDT. For the initial implementation, the MDT would just check for "ROOT" (if it is sent), and return -ENODEV if it is something else, but it would at least get clients ready for the next step.</p> <p>Has this patch been held up just due to the tab/space formatting issues since last January?</p> <p>We are looking at working around this with bind mounts, and I ran across this patch.</p> <p>Scott</p> <p>Scott</p> <p>Thanks for drawing attention to this.</p> <p>Lai</p> <p>Kit has moved on to a new job. Could you please refresh the patch on his behalf?</p> <p>Thanks</p> <p>Peter</p> <p>As Andreas suggested above, this feature should support mount with fileset name instead of "traverse directory on client" hack. This is not a minor change, I'll update it later.</p> <p>Oleg Drokin (oleg.drokin@intel.com) merged in patch <a href="http://review.whamcloud.com/5007/" class="external-link" target="_blank" rel="nofollow noopener">http://review.whamcloud.com/5007/</a><br/> Subject: <a href="https://jira.whamcloud.com/browse/LU-28" title="Ability for client to mount subdirectories of a Lustre filesystem" class="issue-link" data-issue-key="LU-28"><del>LU-28</del></a> fileset: add fileset mount support<br/> Project: fs/lustre-release<br/> Branch: master<br/> Current Patch Set: <br/> Commit: 9bd3e64b817ccbbf9760864b84908b87e13ebfe8</p> <p>The patch has landed to master for 2.9.0</p> Gantt End to End LUDOC-319 Related LU-5989 Bugzilla ID 15267.0 Development Epic metadata mount Rank 1|hzvyh3: Rank (Obsolete) 9769 Severity