找回密码
 注册
查看: 618|回复: 1

记一次模拟测试cephfs reset恢复状态 1 filesystem is degraded

[复制链接]

1

主题

0

回帖

12

积分

管理员

积分
12
QQ
发表于 2022-8-5 14:26:41 | 显示全部楼层 |阅读模式
[root@ceph-3 test]# rados ls -p metedata |grep 2.00000000 |xargs -i rados -p metedata rm {}
/ g. ?8 K+ R7 Z( q) E8 z[root@ceph-3 test]# rados ls -p metedata |grep 2.00000000
* P$ G; z: t: G  {* k& {8 _2 A& j[root@ceph-3 test]# systemctl restart ceph-mds@
0 g" P5 z1 Y) m* q- g; X% Eceph-mds@                ceph-mds@ceph-3.service  - a2 ~. q, X) c$ {1 K
[root@ceph-3 test]# systemctl restart ceph-mds@
8 u1 G/ H  z  Y- s1 M; dceph-mds@                ceph-mds@ceph-3.service  / j5 G. m0 k# f, C5 m! w6 ~
[root@ceph-3 test]# systemctl restart ceph-mds@
. G4 h) s: J* i9 k" d# D9 E" {: W! Q3 Rceph-mds@                ceph-mds@ceph-3.service  
0 C2 s% d6 n' Q. z4 a3 c; N( [- L0 D[root@ceph-3 test]# systemctl restart ceph-mds@ceph-3.service , p! U) x' y$ D6 [$ x3 J
[root@ceph-3 test]# ceph fs status
. t: v5 ^2 T6 {8 y5 [: U                                         #Show the status of a CephFS filesystem
( ?; Q6 }3 l& a$ G& T+ N7 q7 Q+ V5 G[root@ceph-3 test]# ceph fs status 0 H7 U7 Y1 [- G
                                         #Show the status of a CephFS filesystem& |; N' f( V* q# w  D: K
[root@ceph-3 test]# ceph fs status ; J3 d; I& p7 s: k' h8 \6 k
cephfs - 1 clients7 s# w  w, v4 ^0 d2 K
======
  Y3 E$ S: m7 d0 {% \# y3 Q+------+--------+--------+---------------+-------+-------++ y# z* w7 a+ E5 r$ x; p7 D( A
| Rank | State  |  MDS   |    Activity   |  dns  |  inos |
2 n# x, _8 @6 ]  n+ D/ ]+------+--------+--------+---------------+-------+-------+
) U% p  q$ h- @7 |9 L( Y|  0   | active | ceph-1 | Reqs:    0 /s |   11  |   13  |
! R3 }; S5 ?5 ^# N( {8 F+------+--------+--------+---------------+-------+-------+5 |- t) J6 Y+ f) X1 t
+------------+----------+-------+-------+0 b0 B4 K- T# C6 _$ f1 M
|    Pool    |   type   |  used | avail |$ `# p! a. F2 I% H; G! h3 E$ L
+------------+----------+-------+-------+' A0 B. h, e- V% V
|  metedata  | metadata | 5911  |  277G |
+ a8 k8 \+ ~, _# L9 l; J& t' F| cephfsdata |   data   | 44.5k |  277G |& ]) i( K- y) t0 P' E8 T, ~
+------------+----------+-------+-------+* Y2 |5 r' a7 ]- |7 l( D
5 N* z/ r. d( K# k& y
+-------------+
6 D. F+ [0 m2 Y) M| Standby MDS |+ e9 A  }' A% B/ y& _! J
+-------------+6 Q/ \- z: g7 T
|    ceph-2   |" u$ L6 y: {# n
|    ceph-3   |
& z; g* \! S; d+-------------+# ?- E( s2 J5 C6 D1 h: |* d
MDS version: ceph version 12.2.8-52.el7 (3af3ca15b68572a357593c261f95038d02f46201) luminous (stable)
/ G* f* C; V  s  M: }7 H[root@ceph-3 test]# ceph fs status
  R0 g. R7 C, Zcephfs - 1 clients, ?$ d9 |( N- j6 R- M* \
======
9 b' f/ g8 s* o8 P% b( i2 @5 A+------+--------+--------+----------+-------+-------+) {7 k, L4 r% e$ q
| Rank | State  |  MDS   | Activity |  dns  |  inos |7 o6 Y8 v2 j# V6 f4 q* [8 @( q
+------+--------+--------+----------+-------+-------+
& H% S/ k' g" l0 I2 @$ l; P9 O+ W|  0   | replay | ceph-3 |          |    0  |    0  |5 O6 u$ |9 {7 F
+------+--------+--------+----------+-------+-------+4 O; c# |0 b4 y+ b* M( I
+------------+----------+-------+-------+
; X" R$ A" d3 q! ]' W|    Pool    |   type   |  used | avail |1 y. I7 ^( w; ]2 p2 p) h
+------------+----------+-------+-------+
  S* q( K6 M: `! f2 P+ {% m|  metedata  | metadata | 5911  |  277G |4 Q4 p, @4 D. ?' K1 U8 C
| cephfsdata |   data   | 44.5k |  277G |
* r4 b) s* s5 o1 i* q3 l& j4 X3 R+------------+----------+-------+-------+5 \' |  g  R; F  m( v5 f* \
1 F9 c7 t; ^: I( ^9 R4 A
+-------------+7 X6 n* W4 Y9 U9 b# C
| Standby MDS |# e) H5 ~( N! ^! m& w8 R/ A8 Y1 p
+-------------+- F0 y$ Z9 S8 `0 `# J$ G+ N
|    ceph-2   |8 b2 ~1 W8 f! H! s
+-------------+
4 u  J5 A* ?+ ~! h7 pMDS version: ceph version 12.2.8-52.el7 (3af3ca15b68572a357593c261f95038d02f46201) luminous (stable)
& C0 x. [. m5 x, v8 A[root@ceph-3 test]# ceph fs status
# l4 a& I& q  hcephfs - 1 clients
/ C& N. S) b8 N( b8 @5 w======# m4 b' ~0 `6 k/ Y! t
+------+-----------+--------+----------+-------+-------+
# F) ?8 V* T  t, ~  x# B| Rank |   State   |  MDS   | Activity |  dns  |  inos |4 B; g8 c" R& T7 J! M7 E
+------+-----------+--------+----------+-------+-------+7 ^2 f4 k4 x: v: e
|  0   | reconnect | ceph-3 |          |    0  |    0  |
, S, S$ Y2 w: I  y+ ^+------+-----------+--------+----------+-------+-------+' e+ P/ ^$ E2 i4 ]9 J; P! O4 Q
+------------+----------+-------+-------+. q' L: }& F0 G( E5 |( Q
|    Pool    |   type   |  used | avail |3 r, t: Y  W* _9 U; U- i1 K- A+ _
+------------+----------+-------+-------+- J8 q. N; y# l  U! Q& Z
|  metedata  | metadata | 5911  |  277G |
2 p9 e4 O- d& Q  W: A9 ^: t| cephfsdata |   data   | 44.5k |  277G |
/ w) W, j9 W9 }; E. `1 Q) M. B+------------+----------+-------+-------+  {2 Y5 F, Y* z$ ~

; r/ l% u$ G- j; t# {6 n2 k7 E. U3 [+-------------+
4 s! Y, R: ~: f5 n6 l+ ~7 \  S| Standby MDS |
. t$ _; G+ }) y0 r) A9 g# \8 U$ I; x+-------------+, u6 o. r' J- y% z
+-------------+# Q! x/ U: F$ w
MDS version: ceph version 12.2.8-52.el7 (3af3ca15b68572a357593c261f95038d02f46201) luminous (stable)
! e. w$ g. |7 H  K" a% s[root@ceph-3 test]# ceph fs status
, R5 o* m1 H- n: d1 _cephfs - 1 clients% V: X1 P4 \5 w3 y
======
3 ^9 k# Q3 l' R, c" P; _* [/ ]+------+-----------+--------+----------+-------+-------+, t: U0 G4 H$ R; W7 \
| Rank |   State   |  MDS   | Activity |  dns  |  inos |# h# {" ?3 R* c* }  @
+------+-----------+--------+----------+-------+-------+
8 h3 u7 z2 ^6 O# a; e) t|  0   | reconnect | ceph-3 |          |    0  |    0  |8 R7 y' H' a/ M) I! G
+------+-----------+--------+----------+-------+-------+
# t; h1 o3 F* Q+------------+----------+-------+-------+
! S) m. n8 M7 l/ ^. K. B% M|    Pool    |   type   |  used | avail |
6 f* e0 `" z( w: G9 G" H2 E+------------+----------+-------+-------+3 H# D+ T5 {2 x2 u5 y: h1 c* _
|  metedata  | metadata | 5911  |  277G |
1 A( ~/ }9 Y+ i' g, ?3 b7 e| cephfsdata |   data   | 44.5k |  277G |
: S8 d9 ^0 m2 A5 n& S+------------+----------+-------+-------+! s  C% n5 m+ D8 n
% f9 q( {+ F$ Z* g4 X' g" W# F8 {+ |; k
+-------------+5 q0 F$ s/ C0 V6 H! z. c5 K
| Standby MDS |* ~( F+ S# g" W& ~4 p8 m
+-------------+% L9 B+ i9 `; b5 R1 ?$ _
+-------------+
; Y. G: |/ W. ^5 |+ X) [/ }) I4 kMDS version: ceph version 12.2.8-52.el7 (3af3ca15b68572a357593c261f95038d02f46201) luminous (stable), g8 p, ]/ |- Z! U$ H/ S
[root@ceph-3 test]# ceph fs status 1 |3 E) d1 l/ C, Y8 Q. R
cephfs - 1 clients% O6 C2 u/ F" t1 h
======
' {3 i9 t" m2 d$ B: t+------+--------+--------+----------+-------+-------+
2 t  x; S; e8 N9 L# `| Rank | State  |  MDS   | Activity |  dns  |  inos |) U5 ~3 x$ @1 y  g9 }. h7 l/ o
+------+--------+--------+----------+-------+-------+2 J2 _  ?6 I' q
|  0   | rejoin | ceph-3 |          |    1  |    3  |) B! a4 _* T( j$ i& J5 e
+------+--------+--------+----------+-------+-------+% J% u* s3 v$ y$ H6 F9 {. {8 l# \4 m
+------------+----------+-------+-------+
; b) g6 `1 q* p! d|    Pool    |   type   |  used | avail |
% {' _% ^8 Y: r$ R+ M+------------+----------+-------+-------+: y1 \0 n9 D$ R/ P5 a* c- j& O
|  metedata  | metadata | 5911  |  277G |8 w& x) J6 f& t8 }) \
| cephfsdata |   data   | 44.5k |  277G |
  U  c) M6 B8 @. S1 Q+------------+----------+-------+-------+
$ U* d5 U8 g! T- `
( z! L# _( g' Z- [# w4 E+-------------+) L9 T  {( Z4 B$ C7 Z
| Standby MDS |
/ R& ?' s; \$ \+ l# p8 a& S+-------------+
5 \' c5 F# i  }/ j* S4 a; h|    ceph-1   |
# A( q, q) J/ k/ U; t+ P9 h6 q% J+-------------+
! i' O8 I6 ~1 Z0 @MDS version: ceph version 12.2.8-52.el7 (3af3ca15b68572a357593c261f95038d02f46201) luminous (stable)
1 W- q/ D- o% y" p6 [[root@ceph-3 test]# ceph fs status
  s1 ^8 o8 x+ h2 t# X5 ocephfs - 1 clients
* b) J* a* T  s. K2 _( h======6 x: t- _9 T6 E* S" n
+------+--------+--------+----------+-------+-------+
8 H& p# W) p2 N3 {/ c" C| Rank | State  |  MDS   | Activity |  dns  |  inos |
* ^2 g& H+ N% M6 r# m+------+--------+--------+----------+-------+-------+
' C0 Z& T& e/ ]% n4 e; I0 `6 I|  0   | rejoin | ceph-3 |          |    1  |    3  |8 o7 U' L( g4 \% x
+------+--------+--------+----------+-------+-------+2 |/ z$ m9 f" Y( U+ J5 W
+------------+----------+-------+-------+- h& i7 o" S  f7 Y6 @( ~
|    Pool    |   type   |  used | avail |" u3 W5 g# Z& U! w: ]
+------------+----------+-------+-------+, R6 R# k! N* `+ z. j* i3 H, n
|  metedata  | metadata | 5911  |  277G |
# [$ u, h1 n- P8 T* ?| cephfsdata |   data   | 44.5k |  277G |
# D& S- C& _' z$ G# u/ \+------------+----------+-------+-------+" N* ?. t6 V/ _8 F. i. h
8 G" |6 C+ j" _/ m7 v/ R$ M
+-------------+
8 d8 v3 f! W; y: ^' s! t7 f| Standby MDS |# e) A9 B& v7 D6 A  j
+-------------+
7 ], ?, k3 w7 t' m0 W" j|    ceph-1   |: o7 h3 c0 \4 F+ R1 a: v
+-------------+
+ V. H) z3 Z! X1 F+ aMDS version: ceph version 12.2.8-52.el7 (3af3ca15b68572a357593c261f95038d02f46201) luminous (stable)- ?2 n5 V4 G& @0 }
[root@ceph-3 test]# ceph fs status 1 o. ?6 [  K6 n. r5 z' ]  ~' n! ~
cephfs - 1 clients* h( ?, H3 J, g: u7 h. \3 Q
======
& u- V! x, {) R6 y+------+--------+--------+----------+-------+-------+
1 Y% ]* L7 E. k& {| Rank | State  |  MDS   | Activity |  dns  |  inos |
. O6 K2 M; s# d  }9 x1 ^6 ?+------+--------+--------+----------+-------+-------+
; m- |# B- \0 Z+ J|  0   | rejoin | ceph-3 |          |    1  |    3  |
7 T1 ?0 O( S4 d" Y5 v! [. z( Y5 z8 }+------+--------+--------+----------+-------+-------+! U' F9 B( X/ e1 b
+------------+----------+-------+-------+
6 _& [; |$ D0 v: n! k' H|    Pool    |   type   |  used | avail |
5 I8 o2 N( C2 t6 A* w2 x7 p+------------+----------+-------+-------+
7 h; B& S/ \% g4 T/ h# N|  metedata  | metadata | 5911  |  277G |
0 z( g2 N! q. g" {+ D. t# J| cephfsdata |   data   | 44.5k |  277G |: r- R; h" X  i. T
+------------+----------+-------+-------+
0 a" N% D% A  P/ G9 e9 {3 |2 L+ S6 n: D  `9 E; b
+-------------+
/ U: w* ^! |. b5 N, G* [" _| Standby MDS |' o7 T+ p: ~2 ]- u2 d
+-------------+
# |- [$ U5 j, s" D9 Z/ h|    ceph-1   |6 q' I3 m( J& J) Y; |6 T
+-------------+% W+ e3 v/ u4 x9 x* L
MDS version: ceph version 12.2.8-52.el7 (3af3ca15b68572a357593c261f95038d02f46201) luminous (stable)
+ G. ~3 E2 c7 V0 |; o) J[root@ceph-3 test]# ceph fs status   C0 X% @6 j/ `* A: |+ o. q
cephfs - 1 clients5 r) m4 Y; N7 i$ q$ H+ N9 f9 e: j
======. b# d5 w5 m: S& I, b- k  G
+------+--------+--------+---------------+-------+-------+& W3 y6 q" s; `& @5 a5 r' O8 e  o
| Rank | State  |  MDS   |    Activity   |  dns  |  inos |
: @2 E: I# s- l0 Y9 v, _+------+--------+--------+---------------+-------+-------+
- {5 \- |! s2 y|  0   | active | ceph-3 | Reqs:    0 /s |    1  |    3  |
9 R6 G  c2 y/ T( ~* e7 l+------+--------+--------+---------------+-------+-------+
; q- v& S7 W8 M( U+ k( N+------------+----------+-------+-------+/ w. Y+ Q0 y9 s5 w* b5 P8 U5 @2 P
|    Pool    |   type   |  used | avail |. O; S% z# S( @
+------------+----------+-------+-------+7 s2 M2 z( Y# Z# W: `
|  metedata  | metadata | 5911  |  277G |
6 R3 e7 D* J& e! g# j5 Z& Q| cephfsdata |   data   | 44.5k |  277G |8 o* D! S1 H+ C0 T' D
+------------+----------+-------+-------+
9 v8 N9 u+ R( b, l7 M: P" V2 i" H$ F/ z- x& G* n& Y
+-------------+9 p. }- ~* v# l8 I
| Standby MDS |: R( Z5 N7 [) `9 W4 Y4 z
+-------------+# K& j7 F/ y. b$ T
|    ceph-2   |
% b+ o/ J: M6 \+ N4 p|    ceph-1   |0 R3 K. |7 b3 D7 w9 V, o
+-------------+
% l) u" S/ u# _1 \MDS version: ceph version 12.2.8-52.el7 (3af3ca15b68572a357593c261f95038d02f46201) luminous (stable)& e3 m5 F6 }2 Y, l3 \8 r
[root@ceph-3 test]# rados ls -p metedata |grep 2.00000000% Z  k, j/ E5 B
[root@ceph-3 test]# rados ls -p metedata |grep 2.00000000+ F) U8 M; z+ x2 c
[root@ceph-3 test]# ls* V7 d8 i/ [0 m& ]5 R
10000000000.00000000  recoverfile.png
8 a$ X" W+ J" ^' D) a: P& V[root@ceph-3 test]# ( h, P* ^" H3 U. d8 |' D4 [& N
[root@ceph-3 test]#
+ h5 C4 H: O; Z% m[root@ceph-3 test]# cd
. P  d( j4 k' Z4 E; u$ {, w$ p[root@ceph-3 ~]# rados ls -p metedata/ |( n8 j# n, f$ `3 c$ F% U
601.00000000
* w4 T/ G7 u! U' }( y9 T, n600.00000000
/ |4 d4 q6 f( k/ D% K603.00000000! {8 {2 C9 P/ T: T8 a
1.00000000.inode
/ |8 l5 ^. L- s5 t- F  b200.00000000" c- b, H6 x* v; X: m6 Y7 g1 e* r
200.000000017 B1 p! J0 c# o- E- I5 j
606.00000000
9 s, M. u7 @/ Y/ X607.00000000
) U8 ?2 v: @: T7 ~% I% W5 P608.00000000# o) l+ ~+ L6 e+ Z
604.000000004 B; G, _+ {  g# J1 q
500.00000000! d' p" ]$ ]$ h
mds_snaptable
# n% v9 y; I0 S2 D3 H0 n) P# |605.00000000
; I) o$ |2 C8 g1 Z9 }( wmds0_inotable
4 q0 x. R6 H0 v* w) B& d100.000000007 k8 @4 q# H9 I" @) J
mds0_sessionmap' c% O. H6 A; i) \4 E3 `
609.00000000: f, P. o; o" u# `
400.00000000
! B' J( m1 ]7 O8 }100.00000000.inode! p! X, u! ^+ E6 f* ?7 \& p$ _2 [
1.00000000' F3 `* Y+ x8 d% b
[root@ceph-3 ~]# rados ls -p metedata |grep 6024 I+ C  _) \0 o" H& ]7 T0 K& L; \% b
[root@ceph-3 ~]# cd /mnt/^C2 Z- ^, `& k  r
[root@ceph-3 ~]# ceph -s6 X7 H+ t5 n5 [* I% A
  cluster:; n3 V' `% S9 n4 v8 P1 \* `. R- ^
    id:     57bf4711-2218-46af-99d6-9c68ae230ce17 _. \6 u1 g! H0 r  e  P
    health: HEALTH_ERR
( A: {/ J4 k# B  j! ^            1 filesystem is degraded8 k2 \0 {8 y& S8 U' X* Q& U. v
            1 mds daemon damaged
; v2 d% O5 x* t- G6 U* Q5 ~/ Z! H# V7 U& x2 u) I
  services:
1 o& d4 ^' W% {    mon: 3 daemons, quorum ceph-2,ceph-1,ceph-3
3 m3 H! U# ^/ S( S# s; c    mgr: ceph-2(active), standbys: ceph-1, ceph-3
' Z" e  O5 Y3 n3 N; u    mds: cephfs-0/1/1 up , 3 up:standby, 1 damaged9 m& t9 s: H, Z/ f8 e
    osd: 6 osds: 6 up, 6 in; M( y. ]; ]7 b
0 H6 y3 X# q8 |" ^! M6 D
  data:
1 r) `* r% n3 b- K( v8 q    pools:   2 pools, 256 pgs' {8 G4 ^# W. a7 t  K8 S& K, e4 h. o
    objects: 21 objects, 50.4KiB
1 P0 ?! Y. t  K9 q! V, f* ?    usage:   738MiB used, 584GiB / 585GiB avail
! s) ~  y; c6 y0 b+ _    pgs:     256 active+clean
8 ^0 E4 ~# j7 S! X5 i4 K
' H4 K4 F) F8 h# z. \[root@ceph-3 ~]# ls5 s( p2 w! @  `* n+ L9 Z3 L
recovery_cephfs.py
! a- ~7 Z0 |- [[root@ceph-3 ~]# ls
6 |! [0 ]& A6 |4 i+ C4 g; jrecovery_cephfs.py
+ n& k; j8 j% H1 N$ \; W* h[root@ceph-3 ~]#
; b0 {9 n, U, w8 F8 r+ b" n2 Z[root@ceph-3 ~]# ls1 Z9 ^( {2 _8 ~" n
recovery_cephfs.py
, j6 K- d5 s! J! _[root@ceph-3 ~]# df -TH7 v! r% w  w9 ?" i- w; S) o* l: U1 n
6 ~3 f3 v2 m6 s

8 E- J( @2 |5 K# z* Q
$ w) a1 n7 X! u. A4 R8 u' b' }" Q^C
% p% z4 q) ~& G) I[root@ceph-3 ~]# ls
; Y) b6 P. y3 ^; Srecovery_cephfs.py$ p& P& }* W+ F' b1 ~. Z9 i: o3 {
[root@ceph-3 ~]# ls7 N! a0 F# J* V" s
recovery_cephfs.py; }/ I" m1 B7 z  N( m" d4 Q5 Q( N
[root@ceph-3 ~]# ls
/ R* M4 c, _; Drecovery_cephfs.py0 E! {; \2 u# M' ]2 `8 l
[root@ceph-3 ~]# ceph -s
5 m# {4 h; M) J1 {  cluster:
- x1 y) @' s. Y    id:     57bf4711-2218-46af-99d6-9c68ae230ce1
/ ?. U2 ^# d- [" a' f+ Y( d    health: HEALTH_ERR
& q  V$ y0 {& h: s            1 filesystem is degraded
. d1 L; S+ `( _/ ]% J* c. c  k  A            1 mds daemon damaged; g2 ]# J: n* ]) _: E

0 l7 j4 W$ A6 ?# N) j  services:
9 v; e7 N: \/ v4 E( W' v    mon: 3 daemons, quorum ceph-2,ceph-1,ceph-3
! y! O$ B8 r8 }    mgr: ceph-2(active), standbys: ceph-1, ceph-30 i$ D% {4 s9 r- }
    mds: cephfs-0/1/1 up , 3 up:standby, 1 damaged. ^, V; ]2 m' T6 K4 t& U
    osd: 6 osds: 6 up, 6 in
, W- G( d+ B+ q/ h, `" u
$ B$ t, E, _) w" m  data:; }4 Z0 ^) O# l# q
    pools:   2 pools, 256 pgs9 b! ?' i; C. _8 }" }& c' `
    objects: 21 objects, 50.4KiB
, N! Z, H6 j2 N    usage:   738MiB used, 584GiB / 585GiB avail7 V9 G! e) l+ N* |% v- q' ]2 G
    pgs:     256 active+clean$ ^$ s& Z' C) |# B/ l& R- @' U
% q+ w" E) a( C" e
[root@ceph-3 ~]# ceph fs status s
! q* e% W# I" Q" I' ]2 Z  z) q9 G% ~8 h3 j7 v$ p  g8 v
+-------------+  I0 @7 ?; A4 J* w2 T3 J! i/ s
| Standby MDS |
6 t  T0 p7 I1 ?' C" `+-------------+6 Q( q0 K( D! S9 {- G
|    ceph-2   |. B0 g* |2 u6 Q& o9 L
|    ceph-1   |
3 U: r, k- |3 ?  g2 I; X|    ceph-3   |
$ |  e1 c- L' W& ]* D6 J- o+-------------+0 e$ z$ ?% {" p; S2 J( G
MDS version: ceph version 12.2.8-52.el7 (3af3ca15b68572a357593c261f95038d02f46201) luminous (stable)
( L1 x, p2 _3 W2 B[root@ceph-3 ~]# ceph fs status 8 c/ i1 s( q3 K% Q+ E1 _- H% t
cephfs - 0 clients
7 g; S! X8 _% N/ a. @======; t& M) {  [5 b
+------+--------+-----+----------+-----+------+
- w6 q+ d, E- c| Rank | State  | MDS | Activity | dns | inos |
8 x) E* k* a" C+------+--------+-----+----------+-----+------+/ b& i( I5 }# o( t. G
|  0   | failed |     |          |     |      |
# `8 z5 g# i) c2 o+------+--------+-----+----------+-----+------+& a/ `" S1 G  m
+------------+----------+-------+-------+
/ r6 p8 {$ x6 ^5 k/ B5 D. y|    Pool    |   type   |  used | avail |& F8 f' m" G3 V
+------------+----------+-------+-------+
2 H/ I, V. M' w! l" E|  metedata  | metadata | 6023  |  277G |# k2 o1 k8 N5 H* a3 O
| cephfsdata |   data   | 44.5k |  277G |
6 a& R1 Z# w1 o6 ~+------------+----------+-------+-------+* D' S3 a  D2 h6 M

; T6 E4 s  m3 N) ]' F5 i+-------------+
+ B$ N5 V1 S9 ~0 ~| Standby MDS |7 K* f- p5 m- Q/ X# n6 B
+-------------+
$ B7 L; T. |4 A5 K9 R2 f|    ceph-2   |  l# N1 `- e) K  T8 {7 M* T1 o
|    ceph-1   |. f1 N, {- V; A" D! }1 M0 C
|    ceph-3   |
9 G+ R  S" i. q. v# {. W+-------------+
/ K9 b( W3 y7 ?MDS version: ceph version 12.2.8-52.el7 (3af3ca15b68572a357593c261f95038d02f46201) luminous (stable)
) R& R& G) Y) H- e( ~[root@ceph-3 ~]# ceph fs status
+ ]& O0 `: W" Z3 S- P, Xcephfs - 0 clients
4 [' b- }  P, Z. H9 J; L6 e======1 f% w2 E0 h" [% ^6 U
+------+--------+-----+----------+-----+------+! A6 b7 h% s! E: i' }! v' q
| Rank | State  | MDS | Activity | dns | inos |
; m5 E* O$ g3 S" T, c( f+------+--------+-----+----------+-----+------+
8 E% ?2 d( b0 F- W  ~; _|  0   | failed |     |          |     |      |, Z" x% c. {+ k1 G1 D; e8 a
+------+--------+-----+----------+-----+------+3 \1 ?8 S# s1 l
+------------+----------+-------+-------+5 a. u: K7 k4 M9 F8 L
|    Pool    |   type   |  used | avail |
8 Z" h8 V4 d3 v+ ?9 p8 o+------------+----------+-------+-------+" i! L1 v& i9 w0 I
|  metedata  | metadata | 6023  |  277G |1 |1 p5 l! o/ l; T5 K
| cephfsdata |   data   | 44.5k |  277G |3 }1 J: s3 B: \$ b( ]
+------------+----------+-------+-------+$ [& v8 S5 j$ y- l7 n5 S
. R0 z. [; {) f3 q6 `3 M# K3 {% M
+-------------++ h1 n  D* l2 U; ?; [  b- v
| Standby MDS |, F9 Z# L5 O8 G% m) O2 D& o& H
+-------------+
# i4 O; g6 T: O3 o1 B|    ceph-2   |
; n. @- p7 E+ `|    ceph-1   |# v: G, h3 Y8 a4 @1 \; S
|    ceph-3   |
9 d- Q5 Y' \- R4 u8 V+-------------+
4 r8 [9 A8 D& ]$ KMDS version: ceph version 12.2.8-52.el7 (3af3ca15b68572a357593c261f95038d02f46201) luminous (stable)
0 ^$ V/ E# v% M4 P' W3 H[root@ceph-3 ~]# df -TH
) `4 ]& ^4 \) @- k^C
) q- c5 m# P9 `+ _/ N  j/ C( f0 }[root@ceph-3 ~]# ls) _6 f! v, ~* E; q# h6 L  h
recovery_cephfs.py% ~2 b4 C8 I4 }0 b4 l
[root@ceph-3 ~]#
8 @# x: W& w: [7 L1 G% e
  b9 }2 J6 i* M! u; [! O# j[root@ceph-3 ~]# ceph fs ls
+ N0 T9 O7 U. W$ |5 M. Kname: cephfs, metadata pool: metedata, data pools: [cephfsdata ]& g4 p% ?$ z: }- ^) I
[root@ceph-3 ~]# ceph fs reset cephfs -& |) J- C' F/ L9 V9 @3 W' q/ C
--admin-daemon     -f                 -n                 -v                 --watch-info
6 a, i6 m  I# R" W- e--admin-socket     --format           --name             --verbose          --watch-sec
' m: Z4 e/ R& y" G2 ^6 c-c                 -h                 -o                 --version          --watch-warn
# }1 L5 M. K& f9 I/ _--cluster          --help             --out-file         -w                 
, _" H' Z- ~% E/ _0 Z--concise          -i                 -s                 --watch            ( {7 C1 h1 t7 C" |; H, ?" o
--conf             --id               --status           --watch-debug      & v9 K+ C3 o; b7 o* g7 j& s
--connect-timeout  --in-file          --user             --watch-error      * q# Y. O( l9 d( X
[root@ceph-3 ~]# ceph fs reset cephfs -
3 ~! |# T8 F5 a% }: s: _0 U2 B+ _7 T--admin-daemon     -f                 -n                 -v                 --watch-info  e) I7 E- @% L" U/ \3 P* b. A
--admin-socket     --format           --name             --verbose          --watch-sec/ Q3 K' N# K. Q/ m' ]5 P
-c                 -h                 -o                 --version          --watch-warn
" h; g# x- @) K7 o  A% e" H& e--cluster          --help             --out-file         -w                 + W, c" ?8 j/ a* P( ]! ?
--concise          -i                 -s                 --watch            
2 f4 [4 ^8 q5 n# g( S2 u% k--conf             --id               --status           --watch-debug      
9 i+ ~0 g0 h/ |/ O6 K2 C$ L--connect-timeout  --in-file          --user             --watch-error      
3 G# j  J+ J, P( O[root@ceph-3 ~]# ceph fs reset cephfs --yes-i-really-mean-it ) |) {# m6 u3 X* ~$ g' G
                                                    #disaster recovery only: reset to a single-MDS map3 _! Y+ ]. V- \. K
[root@ceph-3 ~]# ceph fs reset cephfs --yes-i-really-mean-it
, d' |7 k- g( n$ o& P& O( p. }[root@ceph-3 ~]# ceph fs ls
. P* i! J# C3 H) h4 Mname: cephfs, metadata pool: metedata, data pools: [cephfsdata ]( C6 u$ W& a- o8 l7 p/ I/ l
[root@ceph-3 ~]# ceph fs status ) @7 G7 N% h4 v/ I3 S1 A- A' `
cephfs - 1 clients2 p( u6 f: q2 b9 P# C, U: q
======  }8 f+ r8 ~3 M1 r) E' s$ W
+------+--------+--------+----------+-------+-------+
+ c+ d/ @6 m% N# V% x. {| Rank | State  |  MDS   | Activity |  dns  |  inos |3 T9 N2 ?1 o, q$ v8 D# h& A
+------+--------+--------+----------+-------+-------+
. l7 L  L9 Y6 n7 B, l) ?. S|  0   | rejoin | ceph-3 |          |    1  |    3  |" y6 c6 D4 o: M1 L1 S  J
+------+--------+--------+----------+-------+-------+# R4 X; H( B0 X0 v9 H0 J& Z
+------------+----------+-------+-------+7 o$ N) w: o8 u
|    Pool    |   type   |  used | avail |: z- A, _; \$ h
+------------+----------+-------+-------+
) Q. R1 a% U) L, p0 U+ x$ A; [|  metedata  | metadata | 6023  |  277G |, ?( K5 b& d4 z0 f6 Z; S
| cephfsdata |   data   | 44.5k |  277G |
* R8 q2 O5 L9 ~7 s: i+------------+----------+-------+-------+' c# v3 u) Q& q4 K) @

  o$ ?! `! f$ C  }" g+-------------+
3 a: z0 _8 Z7 ~7 M' \% ?| Standby MDS |3 d& E4 k# p* ?  T  R8 q' c
+-------------+" ^* B4 m! x$ ?' E9 ?, T. X
|    ceph-2   |
+ r7 W1 m, H% u9 i" Q- _|    ceph-1   |3 }' ]. J* w: k+ M" Q1 g
+-------------+- a! I. G, x; c5 u. L1 i& _
MDS version: ceph version 12.2.8-52.el7 (3af3ca15b68572a357593c261f95038d02f46201) luminous (stable)
0 ~  [( z! h. m  _# w( {5 X[root@ceph-3 ~]# ceph fs status
) G/ p/ b2 n9 a# x; Kcephfs - 1 clients% i1 z& s9 D6 L, n2 T) M* F
======5 J- e1 @1 F, _  p0 d7 O% g& y) }
+------+--------+--------+---------------+-------+-------+; }1 z% {, s# i5 x8 @0 ~; R! r
| Rank | State  |  MDS   |    Activity   |  dns  |  inos |# l% Z* X- {: r0 W& y# z4 g: D2 Z4 R
+------+--------+--------+---------------+-------+-------+
1 a. {  K. r% e) g- S|  0   | active | ceph-3 | Reqs:    0 /s |    1  |    3  |' u9 B% R1 G. x2 M& M% v! q( b1 ]! j6 b
+------+--------+--------+---------------+-------+-------+
* E4 ?( o' n. }* }+ c3 {+------------+----------+-------+-------+
% N) {2 B! i0 x4 n5 Y|    Pool    |   type   |  used | avail |7 N7 e2 z  V) C, C) ]# y0 ?) e
+------------+----------+-------+-------+  k, M* Z0 B# u) Q) ^8 o
|  metedata  | metadata | 6135  |  277G |
( E3 [' A4 L) S: W| cephfsdata |   data   | 44.5k |  277G |0 H+ N3 [( e, r
+------------+----------+-------+-------+
: S) y. K- u* J7 y9 D
( P5 Q. o% Y  @1 t% t/ ^+-------------+: l% x5 ?' W+ Z) S0 Y* X
| Standby MDS |0 Y/ x& v+ C- M( b. z
+-------------+: e7 l: X& o) p4 Y
|    ceph-2   |0 R3 t4 K5 d) u8 X; U6 W0 H
|    ceph-1   |
/ M6 [: d2 L* D7 i' I+-------------+
$ r4 g3 C) f3 b; i/ \1 qMDS version: ceph version 12.2.8-52.el7 (3af3ca15b68572a357593c261f95038d02f46201) luminous (stable)5 U# p; h3 k$ A  l' \
[root@ceph-3 ~]# 9 e- P8 Q* l0 b) w8 U
) I2 V: E8 K0 V; l/ S
: \' N4 x. ^) v6 g/ J4 a

1

主题

0

回帖

12

积分

管理员

积分
12
QQ
 楼主| 发表于 2022-8-11 09:25:06 | 显示全部楼层
解决:7 g; t  _! r0 V- \7 ?- @# Q
针对原因一:6 J3 X7 f$ x- b9 U- a
( D7 o; l1 V9 `" v4 w0 Y
此次情况比较特殊,ceph health detail命令竟然看不到stuck的osd,解决如下:9 T) v' V' b& f* A! B
把 使用率比较高的osd降权重,手动触发osd的数据迁移。0 `% p; R0 `8 _7 R% q  E  U9 E
重新ceph health detail命令,此时会发现stuck的osd,重启相应的osd,mds状态恢复。* i' c' u! v- K; T* {1 _% M
针对原因二(收效甚微):* i) Q1 e& m) E) C. t' u
* d: q  h7 R1 a/ ?: u: i
调小mds mds_cache_memory_limit到40G 以下
% o% i0 }% ^& u: B* V; N# ceph tell mds.\* injectargs '--mds_cache_memory_limit=40000000000'; [  S$ k  I0 b
在/etc/ceph/ceph.conf修改,传到所有ceph节点  \# \# o: x. D, J5 S; K( V7 A
针对原因三(主要原因):8 K. c& q5 c' U( l! u% k

$ V: d: H! h! v# q方法一 mds多活热备调低敏感度,有两个操作:0 x6 d, L( [0 |$ g

# {0 D. z, O' Q8 V调高mds_beacon_grace=300,调高超时时间。
3 g" ?% |; \5 Z. d" r  m& Z  o, h
, P4 k1 B% S$ X- W/ I1 l! D  写到/etc/ceph/ceph.conf里,传到所有mds节点,然后重启mds
) w: F& |; m4 {* A, w+ D3 W' _0 W
mds采用热备模式,替换原来冷备模式,加快mds加入速度。
5 y4 j* {$ O2 i$ x; D, J+ E9 Q' T0 a' e/ D" _  z
  ceph fs set <fs name> allow_standby_replay true2 H+ m5 W) n4 u7 ^# E5 @

* P2 r# T' u+ b+ T方法二 mds多活热备静态目录方法,即给目录划分不同的mds6 X+ b( u7 a0 V9 @  O% s0 B

! L" A/ U$ e1 t8 y命令:
3 f8 A3 ?- L! W# \  }' x. v  z3 `9 gsetfattr -n ceph.dir.pin -v 2 path/to/dir
: ?" T  B2 [  w1 G) @" ~! k9 S6 Y& T  Qmkdir -p a/b1 D) i/ D. }3 i3 `) j4 b
# “a” and “a/b” both start without an export pin set
+ m$ i& r  C7 D- g% jsetfattr -n ceph.dir.pin -v 1 a/
2 q$ S* k7 a' ]$ A3 @# a and b are now pinned to rank 1: P) ]) f& z& a8 W
setfattr -n ceph.dir.pin -v 0 a/b1 [/ m* p) B6 D* F0 [/ \: V# L6 W
# a/b is now pinned to rank 0 and a/ and the rest of its children are still pinned to rank 1
  L) r2 ~( m/ a) }
3 n2 z5 G# y0 H
" N; c+ k, Y0 E4 ?" F2 I5 J1 W
您需要登录后才可以回帖 登录 | 注册

本版积分规则

返回首页|Archiver|手机版|小黑屋|易陆发现技术论坛 ( 蜀ICP备2026014127号-1 )

GMT+8, 2026-6-11 23:50 , Processed in 0.022040 second(s), 23 queries .

Powered by Discuz! X5.0

© 2001-2026 Discuz! Team.

快速回复 返回顶部 返回列表