2 LFSCK: an online file system checker for Lustre
3 ===============================================
5 LFSCK is an online tool to scan, check and repair a Lustre file system that can
6 be used with a file system that is mounted and in use. It contains three main
7 LFSCK components: OI Scrub (primarily of use for ldiskfs-based backend), Layout
8 LFSCK, and Namespace LFSCK. Each component identifies different types of Lustre
11 LFSCK does not verify the on-disk format and assumes that it is consistent. For
12 ldiskfs-based backend, e2fsck from e2fsprogs should be used to ensure the on
13 disk format is consistent. ZFS is designed to always have a valid on-disk
14 structure and as a result, no 'fsck' is necessary.
18 OI scrub is of primary use for ldiskfs-based targets. It maintains the ldiskfs
19 special OI mapping consistency, reconstructs the OI mapping after the target
20 is restored from file-level backup, and upgrades (if necessary) the OI mapping
21 when target (MDT/OST) is upgraded from a previous release.
25 Layout LFSCK is concerned with consistency between metadata targets (MDTs) and
26 object storage targets (OSTs). It automatically corrects inconsistencies where
31 Namespace LFSCK is concerned with consistency across the Lustre namespace.
32 Namespace LFSCK works transparently across single and multiple MDTs.
34 Quick usage instructions
35 ===============================================
39 If you only want OI scrub on a given MDT or OST, use this command on the given
41 # lctl lfsck_start -t scrub -M ${FSNAME}-${TARGETNAME}
43 (FSNAME: the specified file system name created during format, e.g. "testfs".
44 TARGETNAME: the target name in the system, e.g. "MDT0000" or "OST0001".)
46 If you want Layout LFSCK or Namespace LFSCK on a given MDT(s) and OST(s), use
47 this command on the specified MDT:
49 # lctl lfsck_start -t namespace -M ${FSNAME}-${MDTNAME}
51 # lctl lfsck_start -t layout -M ${FSNAME}-${MDTNAME}
53 (MDTNAME: the MDT name in the system, e.g. "MDT0000", "MDT0001".)
55 You can trigger multiple LFSCK components via single LFSCK command:
56 # lctl lfsck_start -t namespace -t layout -M ${FSNAME}-${MDTNAME}
58 For more usage, please run:
61 * review the status of LFSCK
63 Each LFSCK component has its own status interface on a given target.
64 For example, the Namespace LFSCK status on the MDT:
65 # lctl get_param -n mdd.${FSNAME}-${MDTNAME}.lfsck_namespace
67 Or the Layout LFSCK status on the OST:
68 # lctl get_param -n obdfilter.${FSNAME}-${OSTNAME}.lfsck_layout
69 NOTE: Layout LFSCK also works on a OST.
71 (OSTNAME: the OST name in the system, e.g. "OST0000", "OST0001".)
73 Or the OI Scrub status on the MDT/OST:
74 # lctl get_param -n osd-ldiskfs.${FSNAME}-${TARGETNAME}.oi_scrub
78 Run the command on the given MDT/OST:
79 # lctl lfsck_stop -M ${FSNAME}-${MDTNAME}
81 To stop all LFSCK across the system:
82 # lctl lfsck_stop -M ${FSNAME} -A
86 ===============================================
89 * control of scanning rate.
90 * automatic checkpoint recovery of an interrupted scan.
91 * monitoring using proc and lctl interfaces.
92 * maintain OI mapping for ldiskfs-based backend.
93 * reconstruction of the OI mapping after the target (MDT/OST) restored from
95 * generate OI mapping for the MDT upgraded from 1.8.
96 * rebuild OI mapping if some of them lost or crashed.
97 * check and repair kinds of namespace related inconsistent issues:
98 * the FID-in-dirent should be consistent with the FID-in-LMA.
99 * the linkEA should be consistent with related name entries.
100 * Dangling name entry: the name entry exists, but related MDT-object does
102 * Orphan MDT-object: the MDT-object exists, but there is no name entry to
104 * Multiple-referenced name entry: more than one MDT-objects point back to
105 the same name entry, but the name entry only references one of them.
106 * Unmatched name entry and MDT-object pairs: the name entry references the
107 MDT-object that has no linkEA for back-reference or points back to another
108 name entry that does not exist or does not reference the MDT-object.
109 * Unmatched object types: the file type in the name entry does not match the
110 type that is claimed by the MDT-object.
111 * Invalid nlink count: the MDT-object's nlink count does not match the number
112 of name entries that reference such MDT-object.
113 * Invalid name hash for striped directory: the name hash for the name entry
114 on a shard of a striped directory does not match the index stored in the
116 * verify layout consistency between MDT and OST:
117 * MDT-object with dangling reference: the MDT-object1 claims that the
118 OST-object1 is its child stripe, but on the OST, the OST-object1 does not
119 exist, or it is not materialized (so does not recognize the MDT-object1 as
121 * Unmatched referenced MDT-object/OST-object pairs: the MDT-object1 claims
122 that the OST-object1 is its child stripe, but the OST-object1 claims that
123 its parent is the MDT-object2 rather than the MDT-object1. On the MDT,
124 the MDT-object2 does not exist, or not recognize the OST-object1 as its
125 child stripe. An additional case exists where the child index stored in
126 the parent layout information does not match the index information stored
128 * Multiple referenced OST-object: the MDT-object1 claims that the OST-object1
129 is its child stripe, but the OST-object1 claims that its parent is the
130 MDT-object2 rather than the MDT-object1. On the other hand, the MDT-object2
131 recognizes the OST-object1 as its child stripe.
132 * Unreferenced (orphan) OST-object: the OST-object1 claims that the
133 MDT-object1 is its parent, but on the MDT, the MDT-object1 does not exist,
134 or it does not recognize the OST-object1 as its child.
138 ===============================================
140 Information about LFSCK can be found in:
141 /proc/fs/lustre/mdd/${FSNAME}-${MDTNAME}/lfsck_{namespace,layout}
142 /proc/fs/lustre/obdfilter/${FSNAME}-${OSTNAME}/lfsck_layout
143 /proc/fs/lustre/osd-ldiskfs/${FSNAME}-${TARGETNAME}/oi_scrub
146 LFSCK master slave design
147 ===============================================
151 The LFSCK master engine resides on each MDT, and is implemented as a kernel
152 thread in the LFSCK layer. The master engine is responsible for scanning on the
153 MDTs and also controls slave engines on OSTs. Scanning on both MDTs and OSTs
154 occurs in two stages. First-stage scanning will identify and resolve most of
155 inconsistencies. In the second stage, information from the first stage will be
156 used to resolve a remaining set of inconsistencies that had uncertain
157 resolution after only one scan.
160 1. The master engine is started either by the user space command or an
161 excessive number of inconsistency events are detected (defined by
162 osd-ldiskfs.<fsname>-<targetname>.full_scrub_threshold_rate). On starting, the
163 master engine sends RPCs to other MDTs (when necessary) to start other master
164 engines and to related OSTs to start the slave engines.
166 2. The master engine on the MDT scans the MDT local device. Each object is
167 checked for the consistency criteria enumerated in the 'features' section of
170 3. After the MDT completes first-stage system scanning, the master engine sends
171 RPCs to related LFSCK engines on other targets to notify that the first-stage
172 scanning is complete on this MDT. The MDT waits until related targets have
173 completed the first-stage scanning. At this point, the first stage scanning is
174 complete and the second-stage scanning begins.
178 The LFSCK slave engine resides on each OST and is implemented as a kernel
179 thread in the LFSCK layer. This kernel thread drives the first-stage system
182 1. When the slave engine is triggered by the RPC from the master engine in the
183 first-stage scanning, the OST scans the local OST device to generate the
184 in-memory OST orphan object index.
186 2. When the first-stage scanning (for both MDTs and OSTs) is complete a list of
187 non-referenced OST-objects has been accumulated. Only objects that are not
188 accessed during the first stage scan are regarded as potential orphans.
190 3. In the second-stage scanning, the OSTs work to resolve orphan objects in the
191 file system. The OST orphan object index is used as input to the second stage.
192 For each item in the index, the presence of a parent MDT object is verified.
193 Orphan objects will either be relinked to an existing file if found - or moved
194 into a new file in .lustre/lost+found.
196 If multiple MDTs are present, MDTs will check/repair MDT-OST consistency in
197 parallel. To avoid scans of the OST device the slave engine will not begin
198 second-stage system scans until all the master engines complete the first-stage
199 system scan. For each OST there is a single OST orphan object index, regardless
200 of how many MDTs are in the MDT-OST consistency check/repair.
203 Object traversal design reference
204 ===============================================
206 Objects are traversed by LFSCK with two methods: object-table based iteration
207 and namespace based directory traversal.
209 * object-table based iteration
211 The Object Storage Device (OSD) is the abstract layer above a concrete backend
212 file system (i.e. ldiskfs, ZFS, Btrfs, etc.). Each OSD implementation differs
213 internally to support concrete file systems. The object-table based iteration
214 is implemented inside the OSD. It uses the backend special efficient scanning
215 method, such as linear scanning for ldiskfs backend, to scan the local device.
216 Such iteration is presented via the OSD API as a virtual index that contains
217 all the objects that reside on this target.
219 * namespace based directory traversal
221 In addition to object-table based iteration, there are directory based items
222 that need scanning for namespace consistency. For example, FID-in-dirent and
223 LinkEA are directory based features.
225 A naive approach to namespace traversal would be to descend recursively from
226 the file system root. However, this approach will typically generate random IO,
227 which for performance reasons should be minimized. In addition, one must
228 consider operations (i.e. rename) taking place within a directory that is
229 currently being scanned. For these reasons a hybrid approach to scanning is
232 1. LFSCK begins object-table based iteration.
234 2. If a directory is discovered then namespace traversal begins. LFSCK does not
235 descend into sub-directories. LFSCK ignores rename operations during the
236 directory traversal because the subsequent object-table based iteration will
237 guarantee processing of renamed objects. Reading directory blocks is a small
238 fraction of the data needed for the objects they reference. In addition, entries
239 in the directory are typically allocated following the directory object on the
240 disk so for many directories the children objects will already be available
241 because of pre-fetch.
243 3. Process each entry in the directory checking the FID-in-dirent and the FID
244 in the object LMA are consistent. Repair if not. Check also that the linkEA
245 points back to the parent object. Check also that '.' and '..' entries are
248 4. Once all directory entries are exhausted, return to object-table based
253 ===============================================
255 source code: file:/lustre/lfsck/
257 operations manual: https://build.hpdd.intel.com/job/lustre-manual/lastSuccessfulBuild/artifact/lustre_manual.xhtml#dbdoclet.lfsckadmin
259 useful links: http://insidehpc.com/2013/05/02/video-lfsck-online-lustre-file-system-checker/
260 http://www.opensfs.org/wp-content/uploads/2013/04/Zhuravlev_LFSCK.pdf
264 ===============================================
266 OSD - Object storage device. A generic term for a storage device with an
267 interface that extends beyond a block-oriented device interface.
269 FID - File IDentifier. A Lustre file system identifies every file and object
270 with a unique 128-bit ID.
272 OI - Object Index. A table that maps the FID to the object's backend identifier.
273 For ldiskfs-based backend, this table must be regenerated if restored from
276 FID-in-dirent - FID in Directory Entry. To enhance the performance of readdir,
277 the FID of a file is recorded in its directory name entry.
279 LMA - Lustre Metadata Attributes. A record of Lustre specific attributes, for
280 example HSM state, self-FID, and so on.
282 linkEA - Link Extended Attributes. When a file is created or hard-linked the
283 parent directory name and FID are recorded as extended attributes to the file.