X-Git-Url: https://git.whamcloud.com/?a=blobdiff_plain;f=TroubleShootingRecovery.xml;h=b6bb0e5ab2238e941f458a5a164d834d13aab8ef;hb=a2060869a7515e40d358fe366f5b32cf7cb794aa;hp=5e02f4c54ee642e0880eca744471177d10064c17;hpb=0f3e4041a827c924fc95ab3a6f82298c46fbc970;p=doc%2Fmanual.git diff --git a/TroubleShootingRecovery.xml b/TroubleShootingRecovery.xml index 5e02f4c..b6bb0e5 100644 --- a/TroubleShootingRecovery.xml +++ b/TroubleShootingRecovery.xml @@ -1,225 +1,1437 @@ - - - - Troubleshooting Recovery - - This chapter describes what to do if something goes wrong during recovery. It describes: - - - + + + Troubleshooting + Recovery + This chapter describes what to do if something goes wrong during + recovery. It describes: + + + + + - - - + + + + - - - + + + + - - - -
- 27.1 Recovering from Errors or <anchor xml:id="dbdoclet.50438225_marker-1292184" xreflabel=""/>Corruption on a Backing File System - When an OSS, MDS, or MGS server crash occurs, it is not necessary to run e2fsck on the file system. ldiskfs journaling ensures that the file system remains coherent. The backing file systems are never accessed directly from the client, so client crashes are not relevant. - The only time it is REQUIRED that e2fsck be run on a device is when an event causes problems that ldiskfs journaling is unable to handle, such as a hardware device failure or I/O error. If the ldiskfs kernel code detects corruption on the disk, it mounts the file system as read-only to prevent further corruption, but still allows read access to the device. This appears as error "-30" (EROFS) in the syslogs on the server, e.g.: - Dec 29 14:11:32 mookie kernel: LDISKFS-fs error (device sdz): ldiskfs_looku\ -p: unlinked inode 5384166 in dir #145170469 - - Dec 29 14:11:32 mookie kernel: Remounting filesystem read-only - In such a situation, it is normally required that e2fsck only be run on the bad device before placing the device back into service. - In the vast majority of cases, Lustre can cope with any inconsistencies it finds on the disk and between other devices in the file system. - lfsck is rarely required for Lustre operation. - For problem analysis, it is strongly recommended that e2fsck be run under a logger, like script, to record all of the output and changes that are made to the file system in case this information is needed later. - If time permits, it is also a good idea to first run e2fsck in non-fixing mode (-n option) to assess the type and extent of damage to the file system. The drawback is that in this mode, e2fsck does not recover the file system journal, so there may appear to be file system corruption when none really exists. - To address concern about whether corruption is real or only due to the journal not being replayed, you can briefly mount and unmount the ldiskfs filesystem directly on the node with Lustre stopped (NOT via Lustre), using a command similar to: - mount -t ldiskfs /dev/{ostdev} /mnt/ost; umount /mnt/ost - - This causes the journal to be recovered. - The e2fsck utility works well when fixing file system corruption (better than similar file system recovery tools and a primary reason why ldiskfs was chosen over other file systems for Lustre). However, it is often useful to identify the type of damage that has occurred so an ldiskfs expert can make intelligent decisions about what needs fixing, in place of e2fsck. - root# {stop lustre services for this device, if running} -root# script /tmp/e2fsck.sda -Script started, file is /tmp/e2fsck.sda -root# mount -t ldiskfs /dev/sda /mnt/ost -root# umount /mnt/ost -root# e2fsck -fn /dev/sda # don't fix file system, just check for corrupt\ -ion -: -[e2fsck output] -: -root# e2fsck -fp /dev/sda # fix filesystem using "prudent" answers (usually\ - 'y') - - In addition, the e2fsprogs package contains the lfsck tool, which does distributed coherency checking for the Lustre file system after e2fsck has been run. Running lfsck is NOT required in a large majority of cases, at a small risk of having some leaked space in the file system. To avoid a lengthy downtime, it can be run (with care) after Lustre is started. + + + + + + +
+ + <indexterm> + <primary>recovery</primary> + <secondary>corruption of backing ldiskfs file system</secondary> + </indexterm>Recovering from Errors or Corruption on a Backing ldiskfs File + System + When an OSS, MDS, or MGS server crash occurs, it is not necessary to + run e2fsck on the file system. + ldiskfs journaling ensures that the file system remains + consistent over a system crash. The backing file systems are never accessed + directly from the client, so client crashes are not relevant for server + file system consistency. + The only time it is REQUIRED that + e2fsck be run on a device is when an event causes + problems that ldiskfs journaling is unable to handle, such as a hardware + device failure or I/O error. If the ldiskfs kernel code detects corruption + on the disk, it mounts the file system as read-only to prevent further + corruption, but still allows read access to the device. This appears as + error "-30" ( + EROFS) in the syslogs on the server, e.g.: + Dec 29 14:11:32 mookie kernel: LDISKFS-fs error (device sdz): + ldiskfs_lookup: unlinked inode 5384166 in dir #145170469 +Dec 29 14:11:32 mookie kernel: Remounting filesystem read-only + In such a situation, it is normally required that e2fsck only be run + on the bad device before placing the device back into service. + In the vast majority of cases, the Lustre software can cope with any + inconsistencies found on the disk and between other devices in the file + system. + + The legacy offline-LFSCK tool included with e2fsprogs is rarely + required for Lustre file system operation. offline-LFSCK is not to be + confused with LFSCK tool, which is part of Lustre and provides online + consistency checking. + + For problem analysis, it is strongly recommended that + e2fsck be run under a logger, like script, to record all + of the output and changes that are made to the file system in case this + information is needed later. + If time permits, it is also a good idea to first run + e2fsck in non-fixing mode (-n option) to assess the type + and extent of damage to the file system. The drawback is that in this mode, + e2fsck does not recover the file system journal, so there + may appear to be file system corruption when none really exists. + To address concern about whether corruption is real or only due to + the journal not being replayed, you can briefly mount and unmount the + ldiskfs file system directly on the node with the Lustre + file system stopped, using a command similar to: + mount -t ldiskfs /dev/{ostdev} /mnt/ost; umount /mnt/ost + This causes the journal to be recovered. + The + e2fsck utility works well when fixing file system + corruption (better than similar file system recovery tools and a primary + reason why + ldiskfs was chosen over other file systems). However, it + is often useful to identify the type of damage that has occurred so an + ldiskfs expert can make intelligent decisions about what + needs fixing, in place of + e2fsck. + root# {stop lustre services for this device, if running} +root# script /tmp/e2fsck.sda +Script started, file is /tmp/e2fsck.sda +root# mount -t ldiskfs /dev/sda /mnt/ost +root# umount /mnt/ost +root# e2fsck -fn /dev/sda # don't fix file system, just check for corruption +: +[e2fsck output] +: +root# e2fsck -fp /dev/sda # fix errors with prudent answers (usually yes) +
+
+ + <indexterm> + <primary>recovery</primary> + <secondary>corruption of Lustre file system</secondary> + </indexterm>Recovering from Corruption in the Lustre File System + In cases where an ldiskfs MDT or OST becomes corrupt, you need to run + e2fsck to correct the local filesystem consistency, then use + LFSCK to run a distributed check on the file system to + resolve any inconsistencies between the MDTs and OSTs, or among MDTs. + + + Stop the Lustre file system. + + + Run + e2fsck -f on the individual MDT/OST that had + problems to fix any local file system damage. + We recommend running + e2fsck under script, to create a log of changes made + to the file system in case it is needed later. After + e2fsck is run, bring up the file system, if + necessary, to reduce the outage window. + + +
+ + <indexterm> + <primary>recovery</primary> + <secondary>orphaned objects</secondary> + </indexterm>Working with Orphaned Objects + The simplest problem to resolve is that of orphaned objects. When + the LFSCK layout check is run, these objects are linked to new files and + put into + .lustre/lost+found/MDTxxxx + in the Lustre file system + (where MDTxxxx is the index of the MDT on which the orphan was found), + where they can be examined and saved or deleted as necessary. + With Lustre version 2.7 and later, LFSCK will + identify and process orphan objects found on MDTs as well. +
+
+
+ + <indexterm> + <primary>recovery</primary> + <secondary>unavailable OST</secondary> + </indexterm>Recovering from an Unavailable OST + One problem encountered in a Lustre file system environment is when + an OST becomes unavailable due to a network partition, OSS node crash, etc. + When this happens, the OST's clients pause and wait for the OST to become + available again, either on the primary OSS or a failover OSS. When the OST + comes back online, the Lustre file system starts a recovery process to + enable clients to reconnect to the OST. Lustre servers put a limit on the + time they will wait in recovery for clients to reconnect. + During recovery, clients reconnect and replay their requests + serially, in the same order they were done originally. Until a client + receives a confirmation that a given transaction has been written to stable + storage, the client holds on to the transaction, in case it needs to be + replayed. Periodically, a progress message prints to the log, stating + how_many/expected clients have reconnected. If the recovery is aborted, + this log shows how many clients managed to reconnect. When all clients have + completed recovery, or if the recovery timeout is reached, the recovery + period ends and the OST resumes normal request processing. + If some clients fail to replay their requests during the recovery + period, this will not stop the recovery from completing. You may have a + situation where the OST recovers, but some clients are not able to + participate in recovery (e.g. network problems or client failure), so they + are evicted and their requests are not replayed. This would result in any + operations on the evicted clients failing, including in-progress writes, + which would cause cached writes to be lost. This is a normal outcome; the + recovery cannot wait indefinitely, or the file system would be hung any + time a client failed. The lost transactions are an unfortunate result of + the recovery process. + + The failure of client recovery does not indicate or lead to + filesystem corruption. This is a normal event that is handled by the MDT + and OST, and should not result in any inconsistencies between + servers. + + + The version-based recovery (VBR) feature enables a failed client to + be ''skipped'', so remaining clients can replay their requests, resulting + in a more successful recovery from a downed OST. For more information + about the VBR feature, see + (Version-based Recovery). + +
+
+ + <indexterm> + <primary>recovery</primary> + <secondary>oiscrub</secondary> + </indexterm> + <indexterm> + <primary>recovery</primary> + <secondary>LFSCK</secondary> + </indexterm>Checking the file system with LFSCK + LFSCK is an administrative tool introduced in Lustre + software release 2.3 for checking and repair of the attributes specific to a + mounted Lustre file system. It is similar in concept to an offline fsck repair + tool for a local filesystem, but LFSCK is implemented to run as part of the + Lustre file system while the file system is mounted and in use. This allows + consistency of checking and repair by the Lustre software without unnecessary + downtime, and can be run on the largest Lustre file systems with negligible + disruption to normal operations. + Since Lustre software release 2.3, LFSCK can verify + and repair the Object Index (OI) table that is used internally to map + Lustre File Identifiers (FIDs) to MDT internal ldiskfs inode numbers, in + an internal table called the OI Table. An OI Scrub traverses this the IO + Table and makes corrections where necessary. An OI Scrub is required after + restoring from a file-level MDT backup ( + ), or in case the OI Table is + otherwise corrupted. Later phases of LFSCK will add further checks to the + Lustre distributed file system state. + In Lustre software release 2.4, LFSCK namespace + scanning can verify and repair the directory FID-in-Dirent and LinkEA + consistency. + In Lustre software release 2.6, LFSCK layout scanning + can verify and repair MDT-OST file layout inconsistencies. File layout + inconsistencies between MDT-objects and OST-objects that are checked and + corrected include dangling reference, unreferenced OST-objects, mismatched + references and multiple references. + In Lustre software release 2.7, LFSCK layout scanning + is enhanced to support verify and repair inconsistencies between multiple + MDTs. + Control and monitoring of LFSCK is through LFSCK and the + /proc file system interfaces. LFSCK supports three types + of interface: switch interface, status interface, and adjustment interface. + These interfaces are detailed below. +
+ LFSCK switch interface +
+ Manually Starting LFSCK +
+ Description + LFSCK can be started after the MDT is mounted using the + lctl lfsck_start command. +
+
+ Usage +lctl lfsck_start <-M | --device [MDT,OST]_device> \ + [-A | --all] \ + [-c | --create_ostobj on | off] \ + [-C | --create_mdtobj on | off] \ + [-e | --error {continue | abort}] \ + [-h | --help] \ + [-n | --dryrun on | off] \ + [-o | --orphan] \ + [-r | --reset] \ + [-s | --speed ops_per_sec_limit] \ + [-t | --type check_type[,check_type...]] \ + [-w | --window_size size] +
+
+ Options + The various + lfsck_start options are listed and described below. + For a complete list of available options, type + lctl lfsck_start -h. + + + + + + + + + Option + + + + + Description + + + + + + + + + -M | --device + + + + The MDT or OST target to start LFSCK on. + + + + + + -A | --all + + + + Start LFSCK on all + targets on all servers simultaneously. + By default, both layout and namespace + consistency checking and repair are started. + + + + + + -c | --create_ostobj + + + + Create the lost OST-object for + dangling LOV EA, + off(default) or + on. If not specified, then the default + behaviour is to keep the dangling LOV EA there without + creating the lost OST-object. + + + + + + -C | --create_mdtobj + + + + Create the lost MDT-object for + dangling name entry, + off(default) or + on. If not specified, then the default + behaviour is to keep the dangling name entry there without + creating the lost MDT-object. + + + + + + -e | --error + + + + Error handle, + continue(default) or + abort. Specify whether the LFSCK will + stop or not if fails to repair something. If it is not + specified, the saved value (when resuming from checkpoint) + will be used if present. This option cannot be changed + while LFSCK is running. + + + + + + -h | --help + + + + Operating help information. + + + + + + -n | --dryrun + + + + Perform a trial without making any changes. + off(default) or + on. + + + + + + -o | --orphan + + + + Repair orphan OST-objects for layout + LFSCK. + + + + + + -r | --reset + + + + Reset the start position for the object iteration to + the beginning for the specified MDT. By default the + iterator will resume scanning from the last checkpoint + (saved periodically by LFSCK) provided it is + available. + + + + + + -s | --speed + + + + Set the upper speed limit of LFSCK processing in + objects per second. If it is not specified, the saved value + (when resuming from checkpoint) or default value of 0 (0 = + run as fast as possible) is used. Speed can be adjusted + while LFSCK is running with the adjustment + interface. + + + + + + -t | --type + + + + The type of checking/repairing that should be + performed. The new LFSCK framework provides a single + interface for a variety of system consistency + checking/repairing operations including: + Without a specified option, the LFSCK component(s) + which ran last time and did not finish or the component(s) + corresponding to some known system inconsistency, will be + started. Anytime the LFSCK is triggered, the OI scrub will + run automatically, so there is no need to specify + OI_scrub in that case. + + namespace: check and repair + FID-in-Dirent and LinkEA consistency. + Lustre-2.7 enhances + namespace consistency verification under DNE mode. + + layout: check and repair MDT-OST + inconsistency. + + + + + + -w | --window_size + + + + The window size for the async request + pipeline. The LFSCK async request pipeline's input/output + may have quite different processing speeds, and there may + be too many requests in the pipeline as to cause abnormal + memory/network pressure. If not specified, then the default + window size for the async request pipeline is 1024. + + + + + +
+
+
+ Manually Stopping LFSCK +
+ Description + To stop LFSCK when the MDT is mounted, use the + lctl lfsck_stop command. +
+
+ Usage +lctl lfsck_stop <-M | --device [MDT,OST]_device> \ + [-A | --all] \ + [-h | --help] +
+
+ Options + The various + lfsck_stop options are listed and described below. + For a complete list of available options, type + lctl lfsck_stop -h. + + + + + + + + + Option + + + + + Description + + + + + + + + + -M | --device + + + + The MDT or OST target to stop LFSCK on. + + + + + + -A | --all + + + + Stop LFSCK on all targets on all servers + simultaneously. + + + + + + -h | --help + + + + Operating help information. + + + + + +
+
+
+
+ LFSCK status interface +
+ LFSCK status of OI Scrub via + <literal>procfs</literal> +
+ Description + For each LFSCK component there is a dedicated procfs interface + to trace the corresponding LFSCK component status. For OI Scrub, the + interface is the OSD layer procfs interface, named + oi_scrub. To display OI Scrub status, the standard + lctl get_param command is used as shown in the + usage below. +
+
+ Usage + lctl get_param -n osd-ldiskfs.FSNAME-[MDT_target|OST_target].oi_scrub +
+
+ Output + + + + + + + + + Information + + + + + Detail + + + + + + + + General Information + + + + + Name: OI_scrub. + + + OI scrub magic id (an identifier unique to OI + scrub). + + + OI files count. + + + Status: one of the status - + init, + scanning, + completed, + failed, + stopped, + paused, or + crashed. + + + Flags: including - + recreated(OI file(s) is/are + removed/recreated), + inconsistent(restored from + file-level backup), + auto(triggered by non-UI mechanism), + and + upgrade(from Lustre software release + 1.8 IGIF format.) + + + Parameters: OI scrub parameters, like + failout. + + + Time Since Last Completed. + + + Time Since Latest Start. + + + Time Since Last Checkpoint. + + + Latest Start Position: the position for the + latest scrub started from. + + + Last Checkpoint Position. + + + First Failure Position: the position for the + first object to be repaired. + + + Current Position. + + + + + + + Statistics + + + + + + Checked total number of objects + scanned. + + + + Updated total number of objects + repaired. + + + + Failed total number of objects that + failed to be repaired. + + + + No Scrub total number of objects + marked + LDISKFS_STATE_LUSTRE_NOSCRUB and + skipped. + + + + IGIF total number of objects IGIF + scanned. + + + + Prior Updated how many objects have + been repaired which are triggered by parallel + RPC. + + + + Success Count total number of + completed OI_scrub runs on the target. + + + + Run Time how long the scrub has run, + tally from the time of scanning from the beginning of + the specified MDT target, not include the + paused/failure time among checkpoints. + + + + Average Speed calculated by dividing + Checked by + run_time. + + + + Real-Time Speed the speed since last + checkpoint if the OI_scrub is running. + + + + Scanned total number of objects under + /lost+found that have been scanned. + + + + Repaired total number of objects + under /lost+found that have been recovered. + + + + Failed total number of objects under + /lost+found failed to be scanned or failed to be + recovered. + + + + + + + +
+
+
+ LFSCK status of namespace via + <literal>procfs</literal> +
+ Description + The + namespace component is responsible for checks + described in . The + procfs interface for this component is in the + MDD layer, named + lfsck_namespace. To show the status of this + component, + lctl get_param should be used as described in the + usage below. +
+
+ Usage + lctl get_param -n mdd. FSNAME-MDT_target.lfsck_namespace +
+
+ Output + + + + + + + + + Information + + + + + Detail + + + + + + + + General Information + + + + + Name: + lfsck_namespace + + + LFSCK namespace magic. + + + LFSCK namespace version.. + + + Status: one of the status - + init, + scanning-phase1, + scanning-phase2, + completed, + failed, + stopped, + paused, + partial, + co-failed, + co-stopped or + co-paused. + + + Flags: including - + scanned-once(the first cycle + scanning has been completed), + inconsistent(one or more + inconsistent FID-in-Dirent or LinkEA entries that have + been discovered), + upgrade(from Lustre software release + 1.8 IGIF format.) + + + Parameters: including + dryrun, + all_targets, + failout, + broadcast, + orphan, + create_ostobj and + create_mdtobj. + + + Time Since Last Completed. + + + Time Since Latest Start. + + + Time Since Last Checkpoint. + + + Latest Start Position: the position the checking + began most recently. + + + Last Checkpoint Position. + + + First Failure Position: the position for the + first object to be repaired. + + + Current Position. + + + + + + + Statistics + + + + + + Checked Phase1 total number of + objects scanned during + scanning-phase1. + + + + Checked Phase2 total number of + objects scanned during + scanning-phase2. + + + + Updated Phase1 total number of + objects repaired during + scanning-phase1. + + + + Updated Phase2 total number of + objects repaired during + scanning-phase2. + + + + Failed Phase1 total number of objets + that failed to be repaired during + scanning-phase1. + + + + Failed Phase2 total number of objets + that failed to be repaired during + scanning-phase2. + + + + directories total number of + directories scanned. + + + + multiple_linked_checked total number + of multiple-linked objects that have been + scanned. + + + + dirent_repaired total number of + FID-in-dirent entries that have been repaired. + + + + linkea_repaired total number of + linkEA entries that have been repaired. + + + + unknown_inconsistency total number of + undefined inconsistencies found in + scanning-phase2. + + + + unmatched_pairs_repaired total number + of unmatched pairs that have been repaired. + + + + dangling_repaired total number of + dangling name entries that have been + found/repaired. + + + + multi_referenced_repaired total + number of multiple referenced name entries that have + been found/repaired. + + + + bad_file_type_repaired total number + of name entries with bad file type that have been + repaired. + + + + lost_dirent_repaired total number of + lost name entries that have been re-inserted. + + + + striped_dirs_scanned total number of + striped directories (master) that have been + scanned. + + + + striped_dirs_repaired total number of + striped directories (master) that have been + repaired. + + + + striped_dirs_failed total number of + striped directories (master) that have failed to be + verified. + + + + striped_dirs_disabled total number of + striped directories (master) that have been + disabled. + + + + striped_dirs_skipped total number of + striped directories (master) that have been skipped + (for shards verification) because of lost master LMV + EA. + + + + striped_shards_scanned total number + of striped directory shards (slave) that have been + scanned. + + + + striped_shards_repaired total number + of striped directory shards (slave) that have been + repaired. + + + + striped_shards_failed total number of + striped directory shards (slave) that have failed to be + verified. + + + + striped_shards_skipped total number + of striped directory shards (slave) that have been + skipped (for name hash verification) because LFSCK does + not know whether the slave LMV EA is valid or + not. + + + + name_hash_repaired total number of + name entries under striped directory with bad name hash + that have been repaired. + + + + nlinks_repaired total number of + objects with nlink fixed. + + + + mul_linked_repaired total number of + multiple-linked objects that have been repaired. + + + + local_lost_found_scanned total number + of objects under /lost+found that have been + scanned. + + + + local_lost_found_moved total number + of objects under /lost+found that have been moved to + namespace visible directory. + + + + local_lost_found_skipped total number + of objects under /lost+found that have been + skipped. + + + + local_lost_found_failed total number + of objects under /lost+found that have failed to be + processed. + + + + Success Count the total number of + completed LFSCK runs on the target. + + + + Run Time Phase1 the duration of the + LFSCK run during + scanning-phase1. Excluding the time + spent paused between checkpoints. + + + + Run Time Phase2 the duration of the + LFSCK run during + scanning-phase2. Excluding the time + spent paused between checkpoints. + + + + Average Speed Phase1 calculated by + dividing + checked_phase1 by + run_time_phase1. + + + + Average Speed Phase2 calculated by + dividing + checked_phase2 by + run_time_phase1. + + + + Real-Time Speed Phase1 the speed + since the last checkpoint if the LFSCK is running + scanning-phase1. + + + + Real-Time Speed Phase2 the speed + since the last checkpoint if the LFSCK is running + scanning-phase2. + + + + + + + +
+
+
+ LFSCK status of layout via + <literal>procfs</literal> +
+ Description + The + layout component is responsible for checking and + repairing MDT-OST inconsistency. The + procfs interface for this component is in the MDD + layer, named + lfsck_layout, and in the OBD layer, named + lfsck_layout. To show the status of this component + lctl get_param should be used as described in the + usage below. +
+
+ Usage + lctl get_param -n mdd. +FSNAME- +MDT_target.lfsck_layout +lctl get_param -n obdfilter. +FSNAME- +OST_target.lfsck_layout +
+
+ Output + + + + + + + + + Information + + + + + Detail + + + + + + + + General Information + + + + + Name: + lfsck_layout + + + LFSCK namespace magic. + + + LFSCK namespace version.. + + + Status: one of the status - + init, + scanning-phase1, + scanning-phase2, + completed, + failed, + stopped, + paused, + crashed, + partial, + co-failed, + co-stopped, or + co-paused. + + + Flags: including - + scanned-once(the first cycle + scanning has been completed), + inconsistent(one or more MDT-OST + inconsistencies have been discovered), + incomplete(some MDT or OST did not + participate in the LFSCK or failed to finish the LFSCK) + or + crashed_lastid(the lastid files on + the OST crashed and needs to be rebuilt). + + + Parameters: including + dryrun, + all_targets and + failout. + + + Time Since Last Completed. + + + Time Since Latest Start. + + + Time Since Last Checkpoint. + + + Latest Start Position: the position the checking + began most recently. + + + Last Checkpoint Position. + + + First Failure Position: the position for the + first object to be repaired. + + + Current Position. + + + + + + + Statistics + + + + + + Success Count: the total number of + completed LFSCK runs on the target. + + + + Repaired Dangling: total number of + MDT-objects with dangling reference have been repaired + in the scanning-phase1. + + + + Repaired Unmatched Pairs total number + of unmatched MDT and OST-object paris have been + repaired in the scanning-phase1 + + + + Repaired Multiple Referenced total + number of OST-objects with multiple reference have been + repaired in the scanning-phase1. + + + + Repaired Orphan total number of + orphan OST-objects have been repaired in the + scanning-phase2. + + + + Repaired Inconsistent Owner total + number.of OST-objects with incorrect owner information + have been repaired in the scanning-phase1. + + + + Repaired Others total number of.other + inconsistency repaired in the scanning phases. + + + + Skipped Number of skipped + objects. + + + + Failed Phase1 total number of objects + that failed to be repaired during + scanning-phase1. + + + + Failed Phase2 total number of objects + that failed to be repaired during + scanning-phase2. + + + + Checked Phase1 total number of + objects scanned during + scanning-phase1. + + + + Checked Phase2 total number of + objects scanned during + scanning-phase2. + + + + Run Time Phase1 the duration of the + LFSCK run during + scanning-phase1. Excluding the time + spent paused between checkpoints. + + + + Run Time Phase2 the duration of the + LFSCK run during + scanning-phase2. Excluding the time + spent paused between checkpoints. + + + + Average Speed Phase1 calculated by + dividing + checked_phase1 by + run_time_phase1. + + + + Average Speed Phase2 calculated by + dividing + checked_phase2 by + run_time_phase1. + + + + Real-Time Speed Phase1 the speed + since the last checkpoint if the LFSCK is running + scanning-phase1. + + + + Real-Time Speed Phase2 the speed + since the last checkpoint if the LFSCK is running + scanning-phase2. + + + + + + + +
+
-
- 27.2 Recovering from <anchor xml:id="dbdoclet.50438225_marker-1292186" xreflabel=""/>Corruption in the Lustre File System - In cases where the MDS or an OST becomes corrupt, you can run a distributed check on the file system to determine what sort of problems exist. Use lfsck to correct any defects found. - - Stop the Lustre file system. - - Run e2fsck -f on the individual MDS / OST that had problems to fix any local file system damage. - We recommend running e2fsck under script, to create a log of changes made to the file system in case it is needed later. After e2fsck is run, bring up the file system, if necessary, to reduce the outage window. - - Run a full e2fsck of the MDS to create a database for lfsck. You must use the -n option for a mounted file system, otherwise you will corrupt the file system. - e2fsck -n -v --mdsdb /tmp/mdsdb /dev/{mdsdev} - - The mdsdb file can grow fairly large, depending on the number of files in the file system (10 GB or more for millions of files, though the actual file size is larger because the file is sparse). It is quicker to write the file to a local file system due to seeking and small writes. Depending on the number of files, this step can take several hours to complete. - Example - e2fsck -n -v --mdsdb /tmp/mdsdb /dev/sdb -e2fsck 1.39.cfs1 (29-May-2006) -Warning: skipping journal recovery because doing a read-only filesystem che\ -ck. -lustre-MDT0000 contains a file system with errors, check forced. -Pass 1: Checking inodes, blocks, and sizes -MDS: ost_idx 0 max_id 288 -MDS: got 8 bytes = 1 entries in lov_objids -MDS: max_files = 13 -MDS: num_osts = 1 -mds info db file written -Pass 2: Checking directory structure -Pass 3: Checking directory connectivity -Pass 4: Checking reference counts -Pass 5: Checking group summary information -Free blocks count wrong (656160, counted=656058). -Fix? no - -Free inodes count wrong (786419, counted=786036). -Fix? no - -Pass 6: Acquiring information for lfsck -MDS: max_files = 13 -MDS: num_osts = 1 -MDS: 'lustre-MDT0000_UUID' mdt idx 0: compat 0x4 rocomp 0x1 incomp 0x4 -lustre-MDT0000: ******* WARNING: Filesystem still has errors ******* - 13 inodes used (0%) - 2 non-contiguous inodes (15.4%) - # of inodes with ind/dind/tind blocks: 0/0/0 -130272 blocks used (16%) - 0 bad blocks - 1 large file - 296 regular files - 91 directories - 0 character device files - 0 block device files - 0 fifos - 0 links - 0 symbolic links (0 fast symbolic links) - 0 sockets - -------- - 387 files - - - Make this file accessible on all OSTs, either by using a shared file system or copying the file to the OSTs. The pdcp command is useful here. - The pdcp command (installed with pdsh), can be used to copy files to groups of hosts. Pdcp is available here: - http://sourceforge.net/projects/pdsh - - Run a similar e2fsck step on the OSTs. The e2fsck --ostdb command can be run in parallel on all OSTs. - e2fsck -n -v --mdsdb /tmp/mdsdb --ostdb /tmp/{ostNdb} \/dev/{ostNdev} - - The mdsdb file is read-only in this step; a single copy can be shared by all OSTs. - If the OSTs do not have shared file system access to the MDS, a stub mdsdb file, {mdsdb}.mdshdr, is generated. This can be used instead of the full mdsdb file. - Example: - [root@oss161 ~]# e2fsck -n -v --mdsdb /tmp/mdsdb --ostdb \ /tmp/ostdb /dev/\ -sda -e2fsck 1.39.cfs1 (29-May-2006) -Warning: skipping journal recovery because doing a read-only filesystem che\ -ck. -lustre-OST0000 contains a file system with errors, check forced. -Pass 1: Checking inodes, blocks, and sizes -Pass 2: Checking directory structure -Pass 3: Checking directory connectivity -Pass 4: Checking reference counts -Pass 5: Checking group summary information -Free blocks count wrong (989015, counted=817968). -Fix? no - -Free inodes count wrong (262088, counted=261767). -Fix? no - -Pass 6: Acquiring information for lfsck -OST: 'lustre-OST0000_UUID' ost idx 0: compat 0x2 rocomp 0 incomp 0x2 -OST: num files = 321 -OST: last_id = 321 - -lustre-OST0000: ******* WARNING: Filesystem still has errors ******* - - 56 inodes used (0%) - 27 non-contiguous inodes (48.2%) - # of inodes with ind/dind/tind blocks: 13/0/0 -59561 blocks used (5%) - 0 bad blocks - 1 large file - 329 regular files - 39 directories - 0 character device files - 0 block device files - 0 fifos - 0 links - 0 symbolic links (0 fast symbolic links) - 0 sockets - -------- - 368 files - - - - Make the mdsdb file and all ostdb files available on a mounted client and run lfsck to examine the file system. Optionally, correct the defects found by lfsck. - script /root/lfsck.lustre.log -lfsck -n -v --mdsdb /tmp/mdsdb --ostdb /tmp/{ost1db} /tmp/{ost2db} ... /lus\ -tre/mount/point - - Example: - script /root/lfsck.lustre.log -lfsck -n -v --mdsdb /home/mdsdb --ostdb /home/{ost1db} \/mnt/lustre/client/ -MDSDB: /home/mdsdb -OSTDB[0]: /home/ostdb -MOUNTPOINT: /mnt/lustre/client/ -MDS: max_id 288 OST: max_id 321 -lfsck: ost_idx 0: pass1: check for duplicate objects -lfsck: ost_idx 0: pass1 OK (287 files total) -lfsck: ost_idx 0: pass2: check for missing inode objects -lfsck: ost_idx 0: pass2 OK (287 objects) -lfsck: ost_idx 0: pass3: check for orphan objects -[0] uuid lustre-OST0000_UUID -[0] last_id 288 -[0] zero-length orphan objid 1 -lfsck: ost_idx 0: pass3 OK (321 files total) -lfsck: pass4: check for duplicate object references -lfsck: pass4 OK (no duplicates) -lfsck: fixed 0 errors - - By default, lfsck reports errors, but it does not repair any inconsistencies found. lfsck checks for three kinds of inconsistencies: - - Inode exists but has missing objects (dangling inode). This normally happens if there was a problem with an OST. - - - - Inode is missing but OST has unreferenced objects (orphan object). Normally, this happens if there was a problem with the MDS. - - - - Multiple inodes reference the same objects. This can happen if the MDS is corrupted or if the MDS storage is cached and loses some, but not all, writes. - - - - If the file system is in use and being modified while the --mdsdb and --ostdb steps are running, lfsck may report inconsistencies where none exist due to files and objects being created/removed after the database files were collected. Examine the lfsck results closely. You may want to re-run the test. - -
- <anchor xml:id="dbdoclet.50438225_pgfId-1290574" xreflabel=""/>27.2.1 <anchor xml:id="dbdoclet.50438225_13916" xreflabel=""/>Working with Orphaned <anchor xml:id="dbdoclet.50438225_marker-1292187" xreflabel=""/>Objects - The easiest problem to resolve is that of orphaned objects. When the -l option for lfsck is used, these objects are linked to new files and put into lost+found in the Lustre file system, where they can be examined and saved or deleted as necessary. If you are certain the objects are not useful, run lfsck with the -d option to delete orphaned objects and free up any space they are using. - To fix dangling inodes, use lfsck with the -c option to create new, zero-length objects on the OSTs. These files read back with binary zeros for stripes that had objects re-created. Even without lfsck repair, these files can be read by entering: - dd if=/lustre/bad/file of=/new/file bs=4k conv=sync,noerror - - Because it is rarely useful to have files with large holes in them, most users delete these files after reading them (if useful) and/or restoring them from backup. - You cannot write to the holes of such files without having lfsck re-create the objects. Generally, it is easier to delete these files and restore them from backup. - To fix inodes with duplicate objects, use lfsck with the -c option to copy the duplicate object to a new object and assign it to a file. One file will be okay and the duplicate will likely contain garbage. By itself, lfsck cannot tell which file is the usable one. +
+ LFSCK adjustment interface +
+ Rate control +
+ Description + The LFSCK upper speed limit can be changed using + lctl set_param as shown in the usage below. +
+
+ Usage + lctl set_param mdd.${FSNAME}-${MDT_target}.lfsck_speed_limit= +N +lctl set_param obdfilter.${FSNAME}-${OST_target}.lfsck_speed_limit= +N +
+
+ Values + + + + + + + + 0 + + + No speed limit (run at maximum speed.) + + + + + positive integer + + + Maximum number of objects to scan per second. + + + + + +
+
+
+ Auto scrub +
+ Description + The + auto_scrub parameter controls whether OI scrub will + be triggered when an inconsistency is detected during OI lookup. It + can be set as described in the usage and values sections + below. + There is also a + noscrub mount option (see + ) which can be used to + disable automatic OI scrub upon detection of a file-level backup at + mount time. If the + noscrub mount option is specified, + auto_scrub will also be disabled, so OI scrub will + not be triggered when an OI inconsistency is detected. Auto scrub can + be renabled after the mount using the command shown in the usage. + Manually starting LFSCK after mounting provides finer control over + the starting conditions. +
+
+ Usage + lctl set_param osd_ldiskfs.${FSNAME}-${MDT_target}.auto_scrub=N + where + Nis an integer as described below. + Lustre software 2.5 and later supports + -P option that makes the + set_param permanent. +
+
+ Values + + + + + + + + 0 + + + Do not start OI Scrub automatically. + + + + + positive integer + + + Automatically start OI Scrub if inconsistency is + detected during OI lookup. + + + + + +
-
- 27.3 Recovering from an <anchor xml:id="dbdoclet.50438225_marker-1292768" xreflabel=""/>Unavailable OST - One of the most common problems encountered in a Lustre environment is when an OST becomes unavailable, because of a network partition, OSS node crash, etc. When this happens, the OST's clients pause and wait for the OST to become available again, either on the primary OSS or a failover OSS. When the OST comes back online, Lustre starts a recovery process to enable clients to reconnect to the OST. Lustre servers put a limit on the time they will wait in recovery for clients to reconnect. The timeout length is determined by the obd_timeout parameter. - During recovery, clients reconnect and replay their requests serially, in the same order they were done originally. Until a client receives a confirmation that a given transaction has been written to stable storage, the client holds on to the transaction, in case it needs to be replayed. Periodically, a progress message prints to the log, stating how_many/expected clients have reconnected. If the recovery is aborted, this log shows how many clients managed to reconnect. When all clients have completed recovery, or if the recovery timeout is reached, the recovery period ends and the OST resumes normal request processing. - If some clients fail to replay their requests during the recovery period, this will not stop the recovery from completing. You may have a situation where the OST recovers, but some clients are not able to participate in recovery (e.g. network problems or client failure), so they are evicted and their requests are not replayed. This would result in any operations on the evicted clients failing, including in-progress writes, which would cause cached writes to be lost. This is a normal outcome; the recovery cannot wait indefinitely, or the file system would be hung any time a client failed. The lost transactions are an unfortunate result of the recovery process. - The version-based recovery (VBR) feature enables a failed client to be ''skipped'', so remaining clients can replay their requests, resulting in a more successful recovery from a downed OST. For more information about the VBR feature, see (Version-based Recovery).