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-2634 Lustre <p>Short symlinks on MDT filesystems formatted with the "extents" feature appear to be created with the EXT4_EXTENTS_FL in osd-ldiskfs, but that shouldn't be happening. e2fsck considers this a corruption and deletes the symlink.</p> <p>While we have never formatted MDT filesystems with "extents" enabled, some users have done this, or enabled it after formatting, and the MDS should not corrupt such filesystems.</p> MDT needs to be formatted with {{--mkfsoptions=&quot;-O extents&quot;}}. This is not a normal configuration, but has been seen in the field LU-2634

short symlinks on MDT with "extents" have EXT4_EXTENTS_FL set

Technical task LU-1366 Blocker Resolved Fixed Emoly Liu Andreas Dilger LB Thu, 17 Jan 2013 13:46:33 +0000 Fri, 22 Mar 2013 20:18:24 +0000 Mon, 18 Feb 2013 15:47:05 +0000 Lustre 2.4.0 Lustre 2.1.2 Lustre 2.1.5 Lustre 2.4.0 Lustre 2.1.5 0 7 <p>I write a conf-sanity test for this case, but I can't reproduce the failure. No symlink error found by e2fsck if MDT is formatted with "-O extents".<br/> Andreas, could you please have a look if anything wrong in my script? Thanks.</p> <div class="preformatted panel" style="border-width: 1px;"><div class="preformattedContent panelContent"> <pre>test_72() { #LU-2634 #enable "-O extents" to overwrite "^extents" local mdsdev=$(mdsdevname 1) local ostdev=$(ostdevname 1) add ${SINGLEMDS} $(mkfs_opts ${SINGLEMDS} ${mdsdev}) \ --reformat --mkfsoptions=\\\"-O extents\\\" $mdsdev || error "start mds with "-O extents failed"" add ost1 $(mkfs_opts ost1 ${ostdev}) --reformat ${ostdev} start_mgsmds || error "MDT start fail" start_ost || error "OST0 start fail" mount_client $MOUNT || error "Unable to mount client" #create 100 short symlinks local fn=100 mkdir -p $DIR/$tdir createmany -o $DIR/$tdir/$tfile-%d $fn || error "create files failed" echo "create $fn short symlinks" for i in $(seq -w 1 $fn); do ln -s $DIR/$tdir/$tfile-$i $MOUNT/$tfile-$i done #umount umount_client $MOUNT || error "umount client failed" stop_ost || error "stop ost failed" stop_mds || error "stop mds failed" #run e2fsck local cmd="$E2FSCK -fnvd $mdsdev" local rc=0 do_facet ${SINGLEMDS} $cmd || rc=$? echo "$cmd return $rc" [ $rc -gt 0 ] &amp;&amp; error "e2fsck $rc errors found" return $rc } run_test 72 "Short symlink won't cause e2fsck error if MDT is formatted with extents enabled" </pre> </div></div> <p>This is the output on my local machine.</p> <div class="panel" style="border-width: 1px;"><div class="panelContent"> <p>== conf-sanity test 72: Short symlink won't cause e2fsck error if MDT is formatted with extents enabled == 01:53:26 (1358531606)<br/> Loading modules from /root/master/lustre/tests/..<br/> detected 2 online CPUs by sysfs<br/> Force libcfs to create 2 CPU partitions<br/> ../libcfs/libcfs/libcfs options: 'cpu_npartitions=2 cpu_npartitions=2'<br/> debug=-1<br/> subsystem_debug=all -lnet -lnd -pinger<br/> gss/krb5 is not supported<br/> quota/lquota options: 'hash_lqs_cur_bits=3'</p> <p> Permanent disk data:<br/> Target: lustre:MDT0000<br/> Index: 0<br/> Lustre FS: lustre<br/> Mount type: ldiskfs<br/> Flags: 0x65<br/> (MDT MGS first_time update )<br/> Persistent mount opts: user_xattr,errors=remount-ro<br/> Parameters: sys.timeout=20 lov.stripesize=1048576 lov.stripecount=0 mdt.identity_upcall=/root/master/lustre/tests/../utils/l_getidentity</p> <p>formatting backing filesystem ldiskfs on /dev/loop0<br/> target name lustre:MDT0000<br/> 4k blocks 50000<br/> options -I 512 -i 2048 -q -O <font color="red">extents</font>,dirdata,uninit_bg,^extents,dir_nlink,quota,huge_file,flex_bg -E lazy_journal_init -F<br/> mkfs_cmd = mke2fs -j -b 4096 -L lustre:MDT0000 -I 512 -i 2048 -q -O extents,dirdata,uninit_bg,^extents,dir_nlink,quota,huge_file,flex_bg -E lazy_journal_init -F /dev/loop0 50000<br/> Writing CONFIGS/mountdata</p> <p> Permanent disk data:<br/> Target: lustre:OST0000<br/> Index: 0<br/> Lustre FS: lustre<br/> Mount type: ldiskfs<br/> Flags: 0x62<br/> (OST first_time update )<br/> Persistent mount opts: errors=remount-ro<br/> Parameters: mgsnode=10.211.55.7@tcp sys.timeout=20</p> <p>formatting backing filesystem ldiskfs on /dev/loop0<br/> target name lustre:OST0000<br/> 4k blocks 50000<br/> options -I 256 -q -O extents,uninit_bg,dir_nlink,quota,huge_file,flex_bg -G 256 -E lazy_itable_init,resize=4290772992,lazy_journal_init -F<br/> mkfs_cmd = mke2fs -j -b 4096 -L lustre:OST0000 -I 256 -q -O extents,uninit_bg,dir_nlink,quota,huge_file,flex_bg -G 256 -E lazy_itable_init,resize=4290772992,lazy_journal_init -F /dev/loop0 50000<br/> Writing CONFIGS/mountdata<br/> start mds service on centos6-3<br/> Starting mds1: -o loop /tmp/lustre-mdt1 /mnt/mds1<br/> Started lustre-MDT0000<br/> start ost1 service on centos6-3<br/> Starting ost1: -o loop /tmp/lustre-ost1 /mnt/ost1<br/> Started lustre-OST0000<br/> mount lustre on /mnt/lustre.....<br/> Starting client: centos6-3: -o user_xattr,flock centos6-3@tcp:/lustre /mnt/lustre<br/> total: 100 creates in 0.12 seconds: 801.81 creates/second<br/> create 100 short symlinks<br/> umount lustre on /mnt/lustre.....<br/> Stopping client centos6-3 /mnt/lustre (opts<img class="emoticon" src="https://jira.whamcloud.com/images/icons/emoticons/smile.png" height="16" width="16" align="absmiddle" alt="" border="0"/><br/> stop ost1 service on centos6-3<br/> Stopping /mnt/ost1 (opts:-f) on centos6-3<br/> stop mds service on centos6-3<br/> Stopping /mnt/mds1 (opts:-f) on centos6-3<br/> e2fsck 1.42.3.wc3 (15-Aug-2012)<br/> Pass 1: Checking inodes, blocks, and sizes<br/> Pass 2: Checking directory structure<br/> Pass 3: Checking directory connectivity<br/> Pass 4: Checking reference counts<br/> Pass 5: Checking group summary information</p> <p> 348 inodes used (0.35%)<br/> 4 non-contiguous files (1.1%)<br/> 1 non-contiguous directory (0.3%)</p> <ol> <li>of inodes with ind/dind/tind blocks: 0/0/0<br/> 16926 blocks used (33.85%)<br/> 0 bad blocks<br/> 1 large file</li> </ol> <p> 192 regular files<br/> 47 directories<br/> 0 character device files<br/> 0 block device files<br/> 0 fifos<br/> 6 links<br/> 100 symbolic links (100 fast symbolic links)<br/> 0 sockets<br/> --------<br/> 345 files<br/> e2fsck -fnvd /tmp/lustre-mdt1 return 0<br/> Resetting fail_loc on all nodes...done.<br/> PASS 72 (37s)<br/> == conf-sanity test complete, duration 64 sec == 01:54:03 (1358531643)</p> </div></div> <p>Probably because option "extents" is in front of "^extents"?</p> <blockquote><p>-O extents,dirdata,uninit_bg,^extents,dir_nlink,quota,huge_file,flex_bg</p></blockquote> <p>When I use --mkfsoptions=\\\"-O ^extents,extents\\\", the error "Fast symlink xxx has EXTENT_FL set. Clear? no" can be reproduced. </p> <p>Sorry, it seems mkfs.lustre works too hard to clear the "extents" option from the MDT feature list. It could also be enabled with tune2fs after formatting.</p> <p>Since the feature extents is disabled by default in current mkfs.lustre code, the patch only needs to fix the problem in the following two situations:<br/> 1. if the users still format MDT with "-O extents", we will prevent setting EXT4_INODE_EXTENTS flag to the new fast symlink inodes;<br/> 2. if the users disable extents feature but the old data already have extents flag, we will clear EXT4_INODE_EXTENTS for the old fast symlink inodes and write back to disk to avoid e2fsck error.</p> <p>Now the patch can pass my local test for the above two situations. BTW, because the part of patch for situation 1 has avoided to set that extent flag for new inodes, it's hard to reproduce situation 2 in conf-sanity test. I simulate it manually and it works.</p> <p>patch tracking at <a href="http://review.whamcloud.com/5154" class="external-link" target="_blank" rel="nofollow noopener">http://review.whamcloud.com/5154</a></p> <p>Looking at the ext4 code in the kernel, it shows something like:</p> <div class="code panel" style="border-width: 1px;"><div class="codeContent panelContent"> <pre class="code-java"><span class="code-keyword">static</span> <span class="code-object">int</span> ext4_symlink(struct inode *dir, struct dentry *dentry, <span class="code-keyword">const</span> <span class="code-object">char</span> *symname) { <span class="code-keyword">if</span> (l &lt; EXT4_N_BLOCKS * 4) { <span class="code-comment">/* clear the extent format <span class="code-keyword">for</span> fast symlink */</span> ext4_clear_inode_flag(inode, EXT4_INODE_EXTENTS); inode-&gt;i_op = &amp;ext4_fast_symlink_inode_operations; } </pre> </div></div> <p>It appears we should be doing the same thing in our own code:</p> <div class="code panel" style="border-width: 1px;"><div class="codeContent panelContent"> <pre class="code-java"> <span class="code-keyword">static</span> <span class="code-object">int</span> osd_ldiskfs_writelink(struct inode *inode, <span class="code-object">char</span> *buffer, <span class="code-object">int</span> buflen) { + <span class="code-comment">/* clear the extent format <span class="code-keyword">for</span> fast symlink */</span> + ldiskfs_clear_inode_flag(inode, LDISKFS_INODE_EXTENTS); memcpy((<span class="code-object">char</span> *)&amp;LDISKFS_I(inode)-&gt;i_data, (<span class="code-object">char</span> *)buffer, buflen); LDISKFS_I(inode)-&gt;i_disksize = buflen; i_size_write(inode, buflen); inode-&gt;i_sb-&gt;s_op-&gt;dirty_inode(inode); <span class="code-keyword">return</span> 0; } </pre> </div></div> <p>Thanks for Andreas' comment! I saw the code in ext4/ldiskfs_symlink(). Now I know my previous change to ext4/ldiskfs is not a right place. It should be fixed in lustre osd-ldiskfs level not ext4/ldiskfs level.</p> <p>BTW, I test the change above in osd_ldiskfs_writelink() and it works.</p> <p>b2_1 port is at <a href="http://review.whamcloud.com/5458" class="external-link" target="_blank" rel="nofollow noopener">http://review.whamcloud.com/5458</a></p> <p>Landed for 2.4</p> Related LU-2627 Development Rank 1|hzvfjj: Rank (Obsolete) 6163