Uploaded image for project: 'Lustre'
  1. Lustre
  2. LU-11638

lfs fid2path should list open-unlinked files

Details

    • Improvement
    • Resolution: Unresolved
    • Major
    • None
    • Lustre 2.10.5
    • 3
    • 9223372036854775807

    Description

      We delete a large file 15TB (4 stripes), but the objects are not deleting.

      pfe24 ~ # lfs fid2path /nobackupp8 0x360635236:0xbd:0x0
      fid2path: error on FID 0x360635236:0xbd:0x0: No such file or directory
      
      debugfs:  stat <946398>
      Inode: 946398   Type: regular    Mode:  0666   Flags: 0x80000
      Generation: 1914950395    Version: 0x00000030:005b5f49
      User: 12137   Group: 21205   Project:     0   Size: 3857685938176
      File ACL: 0
      Links: 1   Blockcount: 7534584632
      Fragment:  Address: 0    Number: 0    Size: 0
       ctime: 0x5be3096c:00000000 -- Wed Nov  7 07:49:00 2018
       atime: 0x5be1ae0c:00000000 -- Tue Nov  6 07:06:52 2018
       mtime: 0x5be3096c:00000000 -- Wed Nov  7 07:49:00 2018
      crtime: 0x5be1ace3:1d2d1844 -- Tue Nov  6 07:01:55 2018
      Size of extra inode fields: 32
      Extended attributes:
        trusted.lma (64)
        lma: fid=[0x100000000:0xa104e0f:0x0] compat=18 incompat=0
        fid: parent=[0x360635236:0xbd:0x0] stripe=3 stripe_size=1048576 stripe_count=4
      EXTENTS:
      (ETB0):234918342, (ETB1):234918343, (ETB2):234918298, (0-2047):3853318144-3853320191, (2048-4095):3854127104-3854129151, (4096-6143):3854174208-3854176255, (6144-8191):3854239744-3854241791, (8192-10239):3854297088-3854299135, (10240-12287):3854319616-3854321663, (12288-14335):3854405632-3854407679, (14336-16383):3854417920-38544199
      67, (16384-18431):3854473216-3854475263, (18432-20479):3854772224-3854774271, (20480-22527):3854839808-3854841855, (22528-24575):3854944256-3854946303, (24576-26623):3855372288-3855374335, (26624-28671):3855370240-3855372287, (28672-30719):3855405056-3855407103, (30720-34815):3855431680-3855435775, (34816-36863):3855423488-385542553
      5, (36864-38911):3855654912-3855656959, (38912-40959):3855663104-3855665151, (40960-43007):3855669248-3855671295, (43008-45055):3855673344-3855675391, (45056-47103):3855691776-3855693823, (47104-49151):3855702016-3855704063, (49152-53247):3855718400-3855722495, (53248-55295):3855730688-3855732735, (55296-57343):3855753216-3855755263
      , (57344-59391):3855747072-3855749119, (59392-61439):3855765504-3855767551, (61440-65535):3855783936-3855788031, (65536-67583):3854090240-3854092287, (67584-69631):3854106624-3854108671, (69632-71679):3854204928-3854206975, (71680-73727):3854295040-3854297087, (73728-75775):3854374912-3854376959, (75776-77823):3854413824-3854415871,
       (77824-79871):3854419968-3854422015, (79872-81919):3854712832-3854714879, (81920-83967):3854811136-3854813183, (83968-86015):3854845952-3854847999, (86016-88063):3854950400-3854952447, (88064-90111):3855187968-3855190015, (90112-92159):3855368192-3855370239, (92160-94207):3855403008-3855405055, (94208-96255):3855409152-3855411199, 
      (96256-98303):3855417344-3855419391, (98304-100351):3855448064-3855450111, (100352-102399):3855650816-3855652863, (102400-104447):3855665152-3855667199, (104448-106495):3855683584-3855685631, (106496-108543):3855689728-3855691775, (108544-110591):3855708160-3855710207, (110592-112639):3855714304-3855716351, (112640-114687):385572659
      2-3855728639, (114688-116735):3855749120-3855751167, (116736-118783):3855773696-3855775743, (118784-120831):3855767552-3855769599, (120832-124927):3855788032-3855792127, (124928-126975):3855818752-3855820799, (126976-129023):3855831040-3855833087, (129024-131071):3855855616-3855857663, (131072-133119):3853316096-3853318143, (133120-
      135167):3854098432-3854100479, (135168-137215):3854112768-3854114815, (137216-139263):3854178304-3854180351, (139264-141311):3854266368-3854268415, (141312-143359):3854292992-3854295039, (143360-145407):3854315520-3854317567, (145408-147455):3854428160-3854430207, (147456-149503):3854409728-3854411775, (149504-151551):3854452736-385
      4454783, (151552-153599):3854471168-3854473215, (153600-155647):3854755840-3854757887, (155648-157695):3854835712-3854837759, (157696-159743):3854942208-3854944255, (159744-161791):3855101952-3855103999, (161792-163839):3855355904-3855357951, (163840-165887):3855413248-3855415295, (165888-167935):3855441920-3855443967, (167936-16998
      3):3855421440-3855423487, (169984-172031):3855642624-3855644671, (172032-174079):3855661056-3855663103, (174080-176127):3855671296-3855673343, (176128-180223):3855677440-3855681535, (180224-182271):3855695872-3855697919, (182272-184319):3855699968-3855702015, (184320-188415):3855704064-3855708159, (188416-190463):3855710208-38557122
      55, (190464-194559):3855736832-3855740927, (194560-196607):3855728640-3855730687, (196608-198655):3854104576-3854106623, (198656-200703):3854143488-3854145535, (200704-202751):3854192640-3854194687, (202752-204799):3854264320-3854266367, (204800-206847):3854299136-3854301183, (206848-208895):3854385152-3854387199, (208896-210943):38
      54407680-3854409727, (210944-212991):3854411776-3854413823, (212992-215039):3854454784-3854456831, (215040-217087):3854747648-3854749695, (217088-219135):3854813184-3854815231, (219136-221183):3854850048-3854852095, (221184-223231):3854952448-3854954495, (223232-225279):3855351808-3855353855, (225280-227327):3855357952-3855359999, (
      227328-229375):3855396864-3855398911, (229376-231423):3855415296-3855417343, (231424-233471):3855435776-3855437823, (233472-235519):3855648768-3855650815, (235520-237567):3855659008-3855661055, (237568-239615):3855667200-3855669247, (239616-241663):3855675392-3855677439, (241664-243711):3855681536-3855683583, (243712-245759):3855687
      680-3855689727, (245760-247807):3855740928-3855742975, (247808-251903):3855722496-3855726591, (251904-253951):3855734784-3855736831, (253952-255999):3855769600-3855771647, (256000-258047):3855745024-3855747071, (258048-260095):3855759360-3855761407, (260096-262143):3855777792-3855779839, (262144-264191):3855761408-3855763455, (26419
      2-266239):3855757312-3855759359, (266240-270335):3855800320-3855804415, (270336-274431):3855792128-3855796223, (274432-276479):3855835136-3855837183, (276480-278527):3855865856-3855867903, (278528-280575):3855872000-3855874047, (280576-282623):3855880192-3855882239, (282624-284671):3855892480-3855894527, (284672-286719):3855904768-3
      855906815, (286720-288767):3855910912-3855912959, (288768-290815):3855919104-3855921151, (290816-292863):3855951872-3855953919, (292864-294911):3855947776-3855949823, (294912-296959):3855960064-3855962111, (296960-299007):3856050176-3856052223, (299008-301055):3856064512-3856066559, (301056-303103):3856101376-3856103423, (303104-305
      151):3856996352-3856998399, (305152-307199):3857031168-3857033215, (307200-309247):3857041408-3857043455, (309248-311295):3857084416-3857086463, (311296-313343):3857057792-3857059839, (313344-315391):3857078272-3857080319, (315392-317439):3857102848-3857104895, (317440-319487):3857123328-3857125375, (319488-321535):3857129472-385713
      1519, (321536-323583):3857145856-3857147903, (323584-325631):3857184768-3857186815, (325632-327679):3857258496-3857260543, (327680-331775):3855804416-3855808511, (331776-333823):3855812608-3855814655, (333824-335871):3855828992-3855831039, (335872-337919):3855833088-3855835135, (337920-339967):3855869952-3855871999, (339968-342015):
      3855878144-3855880191, (342016-344063):3855902720-3855904767, (344064-348159):3855906816-3855910911, (348160-350207):3855927296-3855929343, (350208-352255):3855953920-3855955967, (352256-354303):3855945728-3855947775, (354304-356351):3855962112-3855964159, (356352-360447):3855984640-3855988735, (360448-362495):3856005120-3856007167,
       (362496-366591):3856041984-3856046079, (366592-368639):3856056320-3856058367, (368640-370687):3856066560-3856068607, (370688-372735):3856459776-3856461823, (372736-374783):3856998400-3857000447, (374784-376831):3857047552-3857049599, (376832-380927):3857074176-3857078271, (380928-382975):3857096704-3857098751, (382976-385023):38571
      41760-3857143807, (385024-387071):3857158144-3857160191, (387072-389119):3857260544-3857262591, (389120-391167):3857270784-3857272831, (391168-393215):3857285120-3857287167, (393216-397311):3855808512-3855812607, (397312-399359):3855796224-3855798271, (399360-401407):3855839232-3855841279, (401408-403455):3855837184-3855839231, (403
      456-405503):3855867904-3855869951, (405504-409599):3855874048-3855878143, (409600-411647):3855898624-3855900671, (411648-413695):3855890432-3855892479, (413696-415743):3855937536-3855939583, (415744-417791):3855925248-3855927295, (417792-419839):3855921152-3855923199, (419840-421887):3855941632-3855943679, (421888-423935):3855970304
      -3855972351, (423936-425983):3855964160-3855966207, (425984-428031):3856054272-3856056319, (428032-430079):3856058368-3856060415, (430080-432127):3856080896-3856082943, (432128-434175):3856099328-3856101375, (434176-436223):3857004544-3857006591, (436224-438271):3857029120-3857031167, (438272-440319):3857049600-3857051647, (440320-4
      42367):3857039360-3857041407, (442368-446463):3857061888-3857065983, (446464-448511):3857080320-3857082367, (448512-450559):3857106944-3857108991, (450560-452607):3857125376-3857127423, (452608-454655):3857139712-3857141759, (454656-456703):3857217536-3857219583, (456704-458751):3857225728-3857227775, (458752-462847):3855886336-3855
      890431, (462848-464895):3855931392-3855933439, (464896-466943):3855917056-3855919103, (466944-468991):3855923200-3855925247, (468992-471039):3855966208-3855968255, (471040-473087):3855958016-3855960063, (473088-475135):3855980544-3855982591, (475136-477183):3856068608-3856070655, (477184-479231):3856072704-3856074751, (479232-481279
      ):3856093184-3856095231, (481280-483327):3856457728-3856459775, (483328-485375):3856967680-3856969727, (485376-487423):3857014784-3857016831, (487424-489471):3857037312-3857039359, (489472-493567):3857043456-3857047551, (493568-495615):3857055744-3857057791, (495616-497663):3857065984-3857068031, (497664-499711):3857082368-385708441
      5, (499712-501759):3857115136-3857117183, (501760-503807):3857133568-3857135615, (503808-505855):3857137664-3857139711, (505856-507903):3857242112-3857244159, (507904-509951):3857272832-3857274879, (509952-511999):3857303552-3857305599, (512000-514047):3857287168-3857289215, (514048-516095):3857336320-3857338367, (516096-518143):385
      7573888-3857575935, (518144-520191):3857790976-3857793023, (520192-522239):3857823744-3857825791, (522240-524287):3857944576-3857946623, (524288-526335):3857268736-3857270783, (526336-528383):3857305600-3857307647, (528384-530431):3857319936-3857321983, (530432-532479):3857330176-3857332223, (532480-534527):3857762304-3857764351, (5
      34528-536575):3857807360-3857809407, (536576-538623):3857811456-3857813503, (538624-540671):3857936384-3857938431, (540672-542719):3857981440-3857983487, (542720-544767):3857985536-3857987583, (544768-546815):3857991680-3857993727, (546816-550911):3858003968-3858008063, (550912-552959):3858032640-3858034687, (552960-555007):38580654
      08-3858067455, (555008-557055):3858079744-3858081791, (557056-559103):3858114560-3858116607, (559104-561151):3858210816-3858212863, (561152-563199):3858214912-3858216959, (563200-565247):3858247680-3858249727, (565248-567295):3858325504-3858327551, (567296-571391):3858335744-3858339839, (571392-573439):3858427904-3858429951, (573440
      -575487):3858497536-3858499583, (575488-577535):3858550784-3858552831, (577536-579583):3858593792-3858595839, (579584-581631):3858575360-3858577407, (581632-583679):3858599936-3858601983, (583680-585727):3858622464-3858624511, (585728-587775):3858632704-3858634751, (587776-589823):3858642944-3858644991, (589824-591871):3857326080-38
      57328127, (591872-593919):3857367040-3857369087, (593920-595967):3857575936-3857577983, (595968-598015):3857797120-3857799167, (598016-600063):3857809408-3857811455, (600064-602111):3857934336-3857936383, (602112-604159):3857946624-3857948671, (604160-606207):3857975296-3857977343, (606208-608255):3858001920-3858003967, (608256-6103
      03):3858022400-3858024447, (610304-612351):3858067456-3858069503, (612352-614399):3858075648-3858077695, (614400-616447):3858087936-3858089983, (616448-618495):3858100224-3858102271, (618496-620543):3858206720-3858208767, (620544-622591):3858204672-3858206719, (622592-624639):3858229248-3858231295, (624640-626687):3858302976-3858305
      023, (626688-628735):3858339840-3858341887, (628736-630783):3858370560-3858372607, (630784-634879):3858528256-3858532351, (634880-636927):3858540544-3858542591, (636928-638975):3858573312-3858575359, (638976-641023):3858577408-3858579455, (641024-643071):3858626560-3858628607, (643072-645119):3858630656-3858632703, (645120-647167):3
      858657280-3858659327, (647168-651263):3858647040-3858651135, (651264-653311):3858704384-3858706431, (653312-655359):3858794496-3858796543, (655360-657407):3857340416-3857342463, (657408-659455):3857551360-3857553407, (659456-661503):3857752064-3857754111, (661504-663551):3857803264-3857805311, (663552-665599):3857854464-3857856511, 
      (665600-667647):3857950720-3857952767, (667648-669695):3857973248-3857975295, (669696-671743):3857987584-3857989631, (671744-673791):3858008064-3858010111, (673792-675839):3858024448-3858026495, (675840-677887):3858069504-3858071551, (677888-679935):3858083840-3858085887, (679936-681983):3858102272-3858104319, (681984-684031):385820
      2624-3858204671, (684032-686079):3858216960-3858219007, (686080-688127):3858286592-3858288639, (688128-690175):3858321408-3858323455, (690176-692223):3858341888-3858343935, (692224-694271):3858358272-3858360319, (694272-696319):3858354176-3858356223, (696320-698367):3858466816-3858468863, (698368-700415):3858544640-3858546687, (7004
      16-702463):3858532352-3858534399, (702464-704511):3858581504-3858583551, (704512-706559):3858610176-3858612223, (706560-708607):3858614272-3858616319, (708608-710655):3858638848-3858640895, (710656-714751):3858659328-3858663423, (714752-716799):3858685952-3858687999, (716800-718847):3858681856-3858683903, (718848-720895):3858710528-
      3858712575, (ETB2):3858664192, (720896-722943):3857989632-3857991679, (722944-724991):3858020352-3858022399, (724992-727039):3858057216-3858059263, (727040-729087):3858071552-3858073599, (729088-731135):3858096128-3858098175, (731136-733183):3858104320-3858106367, (733184-737279):3858225152-3858229247, (737280-739327):3858305024-385
      8307071, (739328-741375):3858352128-3858354175, (741376-743423):3858356224-3858358271, (743424-745471):3858495488-3858497535, (745472-749567):3858536448-3858540543, (749568-751615):3858579456-3858581503, (751616-753663):3858571264-3858573311, (753664-755711):3858606080-3858608127, (755712-757759):3858612224-3858614271, (757760-75980
      7):3858640896-3858642943, (759808-761855):3858644992-3858647039, (ETB2):3858629888, (761856-763903):3858667520-3858669567, (763904-765951):3858679808-3858681855, (765952-767999):3858706432-3858708479, (768000-770047):3858651136-3858653183, (770048-772095):3858735104-3858737151, (772096-774143):3858825216-3858827263, (774144-776191):
      3858849792-3858851839, (776192-778239):3858857984-3858860031, (778240-782335):3858864128-3858868223, (782336-784383):3858872320-3858874367, (784384-786431):3858907136-3858909183, (786432-788479):3858677760-3858679807, (788480-790527):3858683904-3858685951, (790528-792575):3858702336-3858704383, (792576-794623):3858796544-3858798591,
       (794624-796671):3858847744-3858849791, (796672-798719):3858851840-3858853887, (798720-800767):3858886656-3858888703, (800768-802815):3858882560-3858884607, (802816-804863):3858870272-3858872319, (804864-806911):3858900992-3858903039, (806912-808959):3858905088-3858907135, (808960-811007):3858972672-3858974719, (811008-813055):38589
      89056-3858991103, (813056-815103):3858997248-3858999295, (815104-817151):3859013632-3859015679, (817152-819199):3859036160-3859038207, (819200-821247):3859179520-3859181567, (821248-823295):3859191808-3859193855, (823296-825343):3859212288-3859214335, (825344-827391):3859230720-3859232767, (827392-829439):3859271680-3859273727, (829
      440-831487):3859265536-3859267583, (831488-833535):3859283968-3859286015, (833536-835583):3859363840-3859365887, (835584-837631):3859560448-3859562495, (837632-839679):3859599360-3859601407, (839680-841727):3859623936-3859625983, (841728-843775):3859654656-3859656703, (843776-845823):3859695616-3859697663, (845824-847871):3859750912
      -3859752959, (847872-851967):3859781632-3859785727, (851968-854015):3858743296-3858745343, (854016-856063):3858833408-3858835455, (856064-858111):3858855936-3858857983, (858112-860159):3858862080-3858864127, (860160-862207):3858876416-3858878463, (862208-864255):3858868224-3858870271, (864256-866303):3858894848-3858896895, (866304-8
      68351):3858903040-3858905087, (868352-870399):3858941952-3858943999, (870400-872447):3858991104-3858993151, (872448-874495):3858999296-3859001343, (874496-884735):3859003392-3859013631, (884736-886783):3859019776-3859021823, (886784-888831):3859204096-3859206143, (888832-890879):3859208192-3859210239, (890880-892927):3859238912-3859
      240959, (892928-894975):3859261440-3859263487, (894976-897023):3859306496-3859308543, (897024-899071):3859286016-3859288063, (899072-901119):3859353600-3859355647, (901120-903167):3859582976-3859585023, (903168-905215):3859632128-3859634175, (905216-909311):3859666944-3859671039, (909312-911359):3859660800-3859662847, (911360-913407
      ):3859685376-3859687423, (913408-915455):3859681280-3859683327, (915456-917503):3859742720-3859744767, (917504-919551):3858874368-3858876415, (919552-921599):3858884608-3858886655, (921600-923647):3858892800-3858894847, (923648-925695):3858933760-3858935807, (925696-927743):3858993152-3858995199, (927744-929791):3859001344-385900339
      1, (929792-931839):3859015680-3859017727, (931840-935935):3859023872-3859027967, (935936-937983):3859038208-3859040255, (937984-940031):3859189760-3859191807, (940032-942079):3859210240-3859212287, (942080-944127):3859275776-3859277823, (944128-946175):3859273728-3859275775, (946176-948223):3859288064-3859290111, (948224-950271):385
      9402752-3859404799, (950272-952319):3859570688-3859572735, (952320-954367):3859597312-3859599359, (954368-956415):3859601408-3859603455, (956416-958463):3859636224-3859638271, (958464-960511):3859662848-3859664895, (960512-962559):3859683328-3859685375, (962560-964607):3859699712-3859701759, (964608-966655):3859748864-3859750911, (9
      66656-970751):3859789824-3859793919, (970752-972799):3859798016-3859800063, (972800-974847):3859828736-3859830783, (974848-978943):3859808256-3859812351, (978944-980991):3859855360-3859857407, (980992-983039):3859884032-3859886079, (983040-985087):3858978816-3858980863, (985088-987135):3859017728-3859019775, (987136-989183):38590402
      56-3859042303, (989184-991231):3859206144-3859208191, (991232-995327):3859257344-3859261439, (995328-997375):3859263488-3859265535, (997376-999423):3859290112-3859292159, (999424-1001471):3859453952-3859455999, (1001472-1003519):3859591168-3859593215, (1003520-1005567):3859585024-3859587071, (1005568-1007615):3859634176-3859636223, 
      debugfs:  ncheck 946398
      Inode   Pathname
      946398  /O/0/d15/168840719
      
       

      Attachments

        Issue Links

          Activity

            [LU-11638] lfs fid2path should list open-unlinked files

            Joseph Gmitter (jgmitter@whamcloud.com) merged in patch https://review.whamcloud.com/33629/
            Subject: LU-11638 admin: debug space management/missing objects
            Project: doc/manual
            Branch: master
            Current Patch Set:
            Commit: 5e609ea889c9626dfe558e170d78ab74dda42230

            gerrit Gerrit Updater added a comment - Joseph Gmitter (jgmitter@whamcloud.com) merged in patch https://review.whamcloud.com/33629/ Subject: LU-11638 admin: debug space management/missing objects Project: doc/manual Branch: master Current Patch Set: Commit: 5e609ea889c9626dfe558e170d78ab74dda42230

            Andreas Dilger (adilger@whamcloud.com) uploaded a new patch: https://review.whamcloud.com/33629
            Subject: LU-11638 admin: debug space management/missing objects
            Project: doc/manual
            Branch: master
            Current Patch Set: 1
            Commit: 46989d10fe59fa2df6d38885cf53839da9add6d4

            gerrit Gerrit Updater added a comment - Andreas Dilger (adilger@whamcloud.com) uploaded a new patch: https://review.whamcloud.com/33629 Subject: LU-11638 admin: debug space management/missing objects Project: doc/manual Branch: master Current Patch Set: 1 Commit: 46989d10fe59fa2df6d38885cf53839da9add6d4
            adilger Andreas Dilger added a comment - - edited

            Reopening this issue because we don't know what the root cause is. We should be able to handle cleanup of open-unlinked files properly, and/or files that are unlinked from the MDT but held open when the MDS crashed.

            I did a small test and saw that lfs fid2path does not report anything for an open-unlinked file:

            client$ lfs path2fid /myth/tmp/1M
            [0x200003ab4:0x435:0x0]
            client$ sleep 10000 < /myth/tmp/1M &
            [1] 13806
            client$ sudo lfs fid2path /myth [0x200003ab4:0x435:0x0]
            /myth/tmp/1M
            mds# lctl get_param mdt.*.exports.*.open_files
            mdt.myth-MDT0000.exports.192.168.20.159@tcp.open_files=[0x200003ab4:0x435:0x0]
            client$ rm /myth/tmp/1M
            mds# lctl get_param mdt.*.exports.*.open_files
            mdt.myth-MDT0000.exports.192.168.20.159@tcp.open_files=[0x200003ab4:0x435:0x0]
            client$ sudo lfs fid2path /myth [0x200003ab4:0x435:0x0]
            lfs fid2path: cannot find '[0x200003ab4:0x435:0x0]': No such file or directory
            client$ lsof /myth
            COMMAND   PID   USER   FD   TYPE    DEVICE SIZE/OFF               NODE NAME
            sleep   13806 mythtv    0r   REG 35,632494  1048576 144115440203858997 /myth/tmp/1M (deleted)
            

            If this happens again in the future, it would be useful to check the mdt.*.exports.*.open_files list to see if the FID is listed there, which also lists the client NID that is holding it open, and you could use lsof on the client to find and kill the offending process (note that lsof will only report file handles open on the local node).

            It would be useful if lfs fid2path listed these open-unlinked files directly, including the client NID and appends (deleted) as lsof does.

            adilger Andreas Dilger added a comment - - edited Reopening this issue because we don't know what the root cause is. We should be able to handle cleanup of open-unlinked files properly, and/or files that are unlinked from the MDT but held open when the MDS crashed. I did a small test and saw that lfs fid2path does not report anything for an open-unlinked file: client$ lfs path2fid /myth/tmp/1M [0x200003ab4:0x435:0x0] client$ sleep 10000 < /myth/tmp/1M & [1] 13806 client$ sudo lfs fid2path /myth [0x200003ab4:0x435:0x0] /myth/tmp/1M mds# lctl get_param mdt.*.exports.*.open_files mdt.myth-MDT0000.exports.192.168.20.159@tcp.open_files=[0x200003ab4:0x435:0x0] client$ rm /myth/tmp/1M mds# lctl get_param mdt.*.exports.*.open_files mdt.myth-MDT0000.exports.192.168.20.159@tcp.open_files=[0x200003ab4:0x435:0x0] client$ sudo lfs fid2path /myth [0x200003ab4:0x435:0x0] lfs fid2path: cannot find '[0x200003ab4:0x435:0x0]': No such file or directory client$ lsof /myth COMMAND PID USER FD TYPE DEVICE SIZE/OFF NODE NAME sleep 13806 mythtv 0r REG 35,632494 1048576 144115440203858997 /myth/tmp/1M (deleted) If this happens again in the future, it would be useful to check the mdt.*.exports.*.open_files list to see if the FID is listed there, which also lists the client NID that is holding it open, and you could use lsof on the client to find and kill the offending process (note that lsof will only report file handles open on the local node). It would be useful if lfs fid2path listed these open-unlinked files directly, including the client NID and appends (deleted) as lsof does.
            pjones Peter Jones added a comment -

            Good news - thanks!

            pjones Peter Jones added a comment - Good news - thanks!

            I was able to delete the object after umounting OST and remount via ldiskfs.

            Please close this issue.

            mhanafi Mahmoud Hanafi added a comment - I was able to delete the object after umounting OST and remount via ldiskfs. Please close this issue.

            Officially, no. In theory, yes it is possible to do that (I sometimes do concurrent ldiskfs mounts on my home filesystem for debugging purposes), but it is not something we have officially tested. If you have a quiet time and could "expect" a crash, but are happy if it works without problems or interference with jobs, then I'd say it should probably be OK to do that. Otherwise, I'd suggest to wait until there is a scheduled outage window and try it then. At worst it would crash, but that is not much worse than just unmounting the OST (causing IO to block), remount as ldiskfs and remove the object, then remount the OST and go through recovery.

            adilger Andreas Dilger added a comment - Officially, no. In theory, yes it is possible to do that (I sometimes do concurrent ldiskfs mounts on my home filesystem for debugging purposes), but it is not something we have officially tested. If you have a quiet time and could "expect" a crash, but are happy if it works without problems or interference with jobs, then I'd say it should probably be OK to do that. Otherwise, I'd suggest to wait until there is a scheduled outage window and try it then. At worst it would crash, but that is not much worse than just unmounting the OST (causing IO to block), remount as ldiskfs and remove the object, then remount the OST and go through recovery.

            This filesystem was not related to LU-11548. The only that may have happened here is the file was delete while it was still being written to from a different client. And then that write thread was killed.

            Is it safe to mount the ost as ldiskfs and delete the object while the OST is mounted as lustre?

            mhanafi Mahmoud Hanafi added a comment - This filesystem was not related to LU-11548 . The only that may have happened here is the file was delete while it was still being written to from a different client. And then that write thread was killed. Is it safe to mount the ost as ldiskfs and delete the object while the OST is mounted as lustre?

            Is this one of the filesystems related to LU-11548? One of the problems seen in that case was multiple hard links to the same OST object. If that is the case here, it is possible that the "real" object was removed, but a hard linked name was left behind and the space was not freed.

            Another possibility is that the object was migrated to a new MDT and the parent FID on the object was not updated. You might be able to search for large files in the directory (possibly belonging to UID 12137 and/or GID 21205) to see if that is the case?

            lfs find /nobackupp8 -size +3T -print0 | xargs -0 ls -l
            

            I assume there were not any other events around this time (eg. MDS or OSS crash)? That shouldn't cause an unlinked file to leave orphan objects behind, but knowing the circumstances around this file may help.

            The orphan object(s) can be linked to /nobackupp8/.lustre/lost+found/ and/or unlinked by running lctl lfsck_start -A -t layout on the MDS (assuming this is not having other issues at the current time). Alternately, instead of running LFSCK, if the lfs find does not turn up any candidate files, you could mount the OST as type ldiskfs and delete the objects manually.

            adilger Andreas Dilger added a comment - Is this one of the filesystems related to LU-11548 ? One of the problems seen in that case was multiple hard links to the same OST object. If that is the case here, it is possible that the "real" object was removed, but a hard linked name was left behind and the space was not freed. Another possibility is that the object was migrated to a new MDT and the parent FID on the object was not updated. You might be able to search for large files in the directory (possibly belonging to UID 12137 and/or GID 21205) to see if that is the case? lfs find /nobackupp8 -size +3T -print0 | xargs -0 ls -l I assume there were not any other events around this time (eg. MDS or OSS crash)? That shouldn't cause an unlinked file to leave orphan objects behind, but knowing the circumstances around this file may help. The orphan object(s) can be linked to /nobackupp8/.lustre/lost+found/ and/or unlinked by running lctl lfsck_start -A -t layout on the MDS (assuming this is not having other issues at the current time). Alternately, instead of running LFSCK, if the lfs find does not turn up any candidate files, you could mount the OST as type ldiskfs and delete the objects manually.

            People

              adilger Andreas Dilger
              mhanafi Mahmoud Hanafi
              Votes:
              0 Vote for this issue
              Watchers:
              7 Start watching this issue

              Dates

                Created:
                Updated: