找回密码
 注册
查看: 617|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 {}
, Y! U4 L0 x* {# X8 e1 v[root@ceph-3 test]# rados ls -p metedata |grep 2.00000000) s0 A4 G3 }: v% G9 r
[root@ceph-3 test]# systemctl restart ceph-mds@0 x7 P9 R  t5 Y' ]/ b# g4 o5 a1 c
ceph-mds@                ceph-mds@ceph-3.service  
7 Z% t8 @# f2 E% V% ~8 W[root@ceph-3 test]# systemctl restart ceph-mds@; v  \7 e3 Q- V1 a. k7 [$ O
ceph-mds@                ceph-mds@ceph-3.service  
- @' |) c+ f) b- x[root@ceph-3 test]# systemctl restart ceph-mds@
% A: H. b) N6 Y" Z2 Oceph-mds@                ceph-mds@ceph-3.service  + O# K8 n  \, L) N9 f; G' J
[root@ceph-3 test]# systemctl restart ceph-mds@ceph-3.service
% E' d" s1 `+ Y4 P[root@ceph-3 test]# ceph fs status ( w+ `( `  \: l1 x. V% h% z8 ]4 G  [
                                         #Show the status of a CephFS filesystem& g, N$ r1 a) V% J* _9 x
[root@ceph-3 test]# ceph fs status
3 R! q2 l  L' W+ _7 W                                         #Show the status of a CephFS filesystem
2 Y5 u3 ~0 c( i! b+ L$ H[root@ceph-3 test]# ceph fs status + r: N3 u' M0 l* b/ n+ a8 f" ^8 o: ~
cephfs - 1 clients  N* J; [# t$ ?* I# g& t! H
======
. }/ b8 L  }8 A3 R- x+------+--------+--------+---------------+-------+-------+
: {- R/ o9 Y6 |4 V3 S. d| Rank | State  |  MDS   |    Activity   |  dns  |  inos |* Z" t' H  h* }5 }9 \
+------+--------+--------+---------------+-------+-------+
. c0 i( n+ K1 H# N0 t|  0   | active | ceph-1 | Reqs:    0 /s |   11  |   13  |4 j4 W! ?; [0 n  M# g  I+ H
+------+--------+--------+---------------+-------+-------+/ M% H. z0 D7 T' `* C8 I
+------------+----------+-------+-------+
9 T/ A) S6 B& x1 V0 r# I4 {|    Pool    |   type   |  used | avail |
0 ?: |, k' d0 r3 }+------------+----------+-------+-------+/ `7 q) z* P3 B0 L6 j% k  t% k  m
|  metedata  | metadata | 5911  |  277G |
0 c& _/ _" X- L, `% K| cephfsdata |   data   | 44.5k |  277G |
8 s  w# U3 G# Y# g6 D+------------+----------+-------+-------+( D  @1 D$ D/ g3 ^3 H# K
% I/ V' V# P) u( v) j5 N3 D: ~
+-------------+
& z2 m, [( f5 q' }6 v| Standby MDS |, L' U# }% {- v- `  w  d
+-------------+
2 ]+ M5 i* U, c5 L+ ^4 _8 y|    ceph-2   |5 [5 f  Z: Q7 V( h* C
|    ceph-3   |& [! r# k, P6 D7 R* p+ {3 c2 i
+-------------+4 a: M2 s! n* E
MDS version: ceph version 12.2.8-52.el7 (3af3ca15b68572a357593c261f95038d02f46201) luminous (stable)5 g' N# _) B' J3 S
[root@ceph-3 test]# ceph fs status
& \) [+ A5 {" ccephfs - 1 clients
# O% K, s# W/ J  E# S======8 z2 c7 d, S1 M: H! @+ E
+------+--------+--------+----------+-------+-------+
& y  @/ e$ L# l# K9 j& G| Rank | State  |  MDS   | Activity |  dns  |  inos |
  ?7 C1 z9 O, {0 I4 ~+------+--------+--------+----------+-------+-------+: K/ ?3 p3 R7 c) j2 _) C2 D+ w
|  0   | replay | ceph-3 |          |    0  |    0  |" p. }% b, I5 T- e& Q$ E' s9 @
+------+--------+--------+----------+-------+-------+
: I/ D7 p1 M/ i% x! A" `+------------+----------+-------+-------++ K6 E9 M/ A+ Z$ c" ^
|    Pool    |   type   |  used | avail |/ |. `; D, ]- H# E
+------------+----------+-------+-------+
+ V7 r- }7 d' b|  metedata  | metadata | 5911  |  277G |1 G$ ^3 [2 r/ B- N  W4 r) s
| cephfsdata |   data   | 44.5k |  277G |
! B8 `- p7 [# A8 A+------------+----------+-------+-------+) b3 m( m/ j! ~! i+ T1 ]) z3 `0 j, T

' y+ }5 G+ c; W+-------------+. `. r: r1 B3 B' w, Y* b
| Standby MDS |/ f# w* F' V" `
+-------------+- `& p* O7 I- g6 m
|    ceph-2   |6 [/ W; s% k7 Z5 F
+-------------+
5 Z9 T! @$ \( q" Q% vMDS version: ceph version 12.2.8-52.el7 (3af3ca15b68572a357593c261f95038d02f46201) luminous (stable)% b  c- u5 c4 o1 k& f" g: o
[root@ceph-3 test]# ceph fs status
7 p: F, j0 H" M; s7 V3 `cephfs - 1 clients5 j* C7 ^! K' ?( _4 D+ S; \. x# }
======9 {1 Z# {; {! T- R! G, ~
+------+-----------+--------+----------+-------+-------+
# v5 `4 X& i5 C4 {| Rank |   State   |  MDS   | Activity |  dns  |  inos |
, }# {# H! p$ i$ \( f: r# |1 X1 L+------+-----------+--------+----------+-------+-------+
6 L' \5 x& M8 P; W|  0   | reconnect | ceph-3 |          |    0  |    0  |, _, t' o5 Y, h* }5 m& T% _2 ~
+------+-----------+--------+----------+-------+-------+( N) _0 j0 m4 l3 a
+------------+----------+-------+-------+
! o  b! }" H- \$ O* s|    Pool    |   type   |  used | avail |" r; L9 Q4 q! h% T  z
+------------+----------+-------+-------+- G! E0 c& [6 n( Q, P1 l
|  metedata  | metadata | 5911  |  277G |  G) T9 W# t/ B; \
| cephfsdata |   data   | 44.5k |  277G |
* V  P& }4 t: f( R! S+------------+----------+-------+-------+
0 M% l% Q9 b: s# T6 M0 H. T- g' U" y7 z" ^+ u7 N1 m
+-------------+  v7 C( Z* D( }6 b0 |5 p! L
| Standby MDS |
5 B2 G: v- f6 K. T7 B7 T0 J+-------------+7 f; F% K. C' U1 @' X4 m9 l
+-------------+9 G6 ~$ {6 W) K6 C) \8 _$ h
MDS version: ceph version 12.2.8-52.el7 (3af3ca15b68572a357593c261f95038d02f46201) luminous (stable)2 r  ]( u3 o7 z! x
[root@ceph-3 test]# ceph fs status ( Z% d4 l  x! ^
cephfs - 1 clients9 l6 Q( Y' i9 y% |) q8 Z: }
======% l8 |% l' `% n8 Y/ S# u
+------+-----------+--------+----------+-------+-------+& R/ @; u# w. u) [
| Rank |   State   |  MDS   | Activity |  dns  |  inos |
" z$ h# l3 S* U$ Q2 y+ O6 ?/ v2 C+------+-----------+--------+----------+-------+-------+
  }9 a. Z4 Y0 {7 G0 e6 [|  0   | reconnect | ceph-3 |          |    0  |    0  |
, {8 T) R% X. p; H0 a+------+-----------+--------+----------+-------+-------+7 G9 w) s+ V2 |3 A. U- _
+------------+----------+-------+-------+
' ]& R$ U3 _- Z) g& w, e|    Pool    |   type   |  used | avail |2 |7 V  _4 |2 ~5 ^3 S
+------------+----------+-------+-------+, K) S) B- k% e/ l' ^& Q5 U  O3 H
|  metedata  | metadata | 5911  |  277G |
9 j8 S! j# X: H5 O6 V3 l" ~# [| cephfsdata |   data   | 44.5k |  277G |
6 Y7 e5 E3 P; O1 l8 E+------------+----------+-------+-------+
" B8 w% ~2 |$ a1 I) C, f8 f% |8 b; H5 A
+-------------+
* E) o8 ^/ @% E3 L+ B! p) X+ w| Standby MDS |
! K7 F; ^3 C3 y8 g+-------------+5 H' @- W% @2 g2 F) v
+-------------+& Y0 L- ^3 L  x0 {4 o7 P  o
MDS version: ceph version 12.2.8-52.el7 (3af3ca15b68572a357593c261f95038d02f46201) luminous (stable)
8 w/ e1 e) d- W' n; y- f5 o6 Y3 p[root@ceph-3 test]# ceph fs status
8 Z' x1 \% I( ocephfs - 1 clients* ?- i8 r5 B7 o$ E' Q
======6 V' R7 t! B: _. F6 t
+------+--------+--------+----------+-------+-------+7 I* @4 h9 O/ G$ W
| Rank | State  |  MDS   | Activity |  dns  |  inos |% H9 n' F) c1 U& v, C  e. {4 Q! x
+------+--------+--------+----------+-------+-------+5 h+ x) x$ Y2 H+ J7 B" c1 I+ k. ]
|  0   | rejoin | ceph-3 |          |    1  |    3  |% m% E. G8 e  w0 j
+------+--------+--------+----------+-------+-------+& D. q& S" d7 j) P  y: Z' E- X2 N
+------------+----------+-------+-------+
; l3 ]& |  C; Y7 p|    Pool    |   type   |  used | avail |$ E& M( ~8 n: l8 `2 ~9 g
+------------+----------+-------+-------+
: i# w# u" Z) M% y3 {|  metedata  | metadata | 5911  |  277G |
+ K( |1 x- A3 u) E! b8 c& Y| cephfsdata |   data   | 44.5k |  277G |
/ E7 Q# f9 n0 C+------------+----------+-------+-------+
& C. N$ I% I, g! B" Z' }  x/ A. ?& \" y  G
+-------------+: |9 U( x. l% ]# p) Y
| Standby MDS |# G2 L/ w6 \! @) S3 |" d# G
+-------------+
) Z0 G4 E2 @6 [4 ^+ `+ Y' h|    ceph-1   |% h3 E" I: L/ M1 ?+ y/ Y
+-------------+$ a! z" a( z5 }# c
MDS version: ceph version 12.2.8-52.el7 (3af3ca15b68572a357593c261f95038d02f46201) luminous (stable)
/ A, [7 M+ P/ |2 K) t[root@ceph-3 test]# ceph fs status ( Z, k/ q4 K/ X  _( |+ s( S
cephfs - 1 clients& Y$ L& G  Q& K8 T. ^: T
======6 e7 ~9 ^# ~5 Q6 l8 A# W* Q: R
+------+--------+--------+----------+-------+-------+
: }# l- Z& [! l. t+ u| Rank | State  |  MDS   | Activity |  dns  |  inos |
+ |  S9 t$ K& V+ a7 y, K" t: n+------+--------+--------+----------+-------+-------+
+ G$ ]* c% j* L* W; X|  0   | rejoin | ceph-3 |          |    1  |    3  |( z+ N- Z0 ?  T) p. N7 |
+------+--------+--------+----------+-------+-------++ A  m( B+ L& W, c- Z* x
+------------+----------+-------+-------+# _2 T1 r( J5 a$ X" i+ D* M8 ~
|    Pool    |   type   |  used | avail |
  e0 }# T3 Z- G3 p7 r+ |+------------+----------+-------+-------+8 ^5 U8 f+ M2 x7 w  I" L6 a
|  metedata  | metadata | 5911  |  277G |# @* p* T  N' H5 `
| cephfsdata |   data   | 44.5k |  277G |% q9 ^* V7 K/ _+ M' h1 r
+------------+----------+-------+-------+
8 P+ R9 u4 {, H. w
* T6 `  l8 t, j* i! U. w7 s5 X+-------------+
/ B/ S3 z4 d* p9 A4 c4 S| Standby MDS |, v+ o9 o0 h  I0 T7 w: G* S" q1 J
+-------------+' B- _/ w0 _2 i2 P" t: @" p7 Y1 E
|    ceph-1   |
. b3 D, q: V& H8 H' J! L+-------------++ [7 V4 o( m" }0 J1 R! C( W# h
MDS version: ceph version 12.2.8-52.el7 (3af3ca15b68572a357593c261f95038d02f46201) luminous (stable)$ G% z. o( E+ v2 x- h
[root@ceph-3 test]# ceph fs status - D7 h' s$ l# r- J8 K0 y' e9 Z# Q
cephfs - 1 clients
, t3 M8 U% p% t, R0 Q& k6 h0 Y, f======
+ `# R) Y1 `% k, R/ b+------+--------+--------+----------+-------+-------+
4 J  \$ Q6 d. r. K- N6 e| Rank | State  |  MDS   | Activity |  dns  |  inos |9 @- |- ^3 w2 Q
+------+--------+--------+----------+-------+-------+
4 C# U1 O2 j: o( N5 j) ~+ e|  0   | rejoin | ceph-3 |          |    1  |    3  |
/ C6 B8 P* ^: `4 q1 I3 U+------+--------+--------+----------+-------+-------+3 I5 b- F  }" X5 e1 d3 s7 e
+------------+----------+-------+-------+$ d9 ^3 }- ?2 a) H0 A) N$ ^
|    Pool    |   type   |  used | avail |6 H" l/ G7 R, g% L0 z
+------------+----------+-------+-------+, B9 `0 `3 [3 p1 q7 ]% Y+ ]! B
|  metedata  | metadata | 5911  |  277G |8 [6 S( \8 T, I( h# a$ Z4 r' c
| cephfsdata |   data   | 44.5k |  277G |; b; |" H/ l, N: r
+------------+----------+-------+-------+" Z! b1 h8 [& `' x, b6 ^
- p5 K1 K! v' w) b
+-------------+) |$ c9 L4 R$ P3 t
| Standby MDS |$ N3 G- D& D. w$ q$ d7 T" _
+-------------+
, [& Y: s. b2 r2 K* V|    ceph-1   |" [! S: A' h; ]' L$ B( z: l5 V) L
+-------------+
2 d. D! r6 F- w! @# HMDS version: ceph version 12.2.8-52.el7 (3af3ca15b68572a357593c261f95038d02f46201) luminous (stable)
! [+ V/ x' a0 [1 F" G' x6 {[root@ceph-3 test]# ceph fs status & F! j0 t4 B' P2 {8 J: ~
cephfs - 1 clients
- O# _! C! ~% |% [======
8 m! z- f( r# F* k& I* F+------+--------+--------+---------------+-------+-------+. Z- o- [" T! ^$ v: I$ F+ C
| Rank | State  |  MDS   |    Activity   |  dns  |  inos |$ b9 N+ C! W5 Y$ P8 g  n
+------+--------+--------+---------------+-------+-------+
+ D+ ?* q* ]0 {- }0 Y|  0   | active | ceph-3 | Reqs:    0 /s |    1  |    3  |
( k% p0 |4 G1 V$ W8 A/ `+------+--------+--------+---------------+-------+-------+
$ R7 `, j- W0 r5 I' M4 v- B! G, R+------------+----------+-------+-------+9 i, V5 Q" {) z9 @
|    Pool    |   type   |  used | avail |9 D; ?! I& Z  B9 O, x! M
+------------+----------+-------+-------+# }5 F# h( O: f1 ~# ^& I! J
|  metedata  | metadata | 5911  |  277G |7 @0 j* f5 F* V0 L) M1 I! X& M5 S
| cephfsdata |   data   | 44.5k |  277G |
' [) R$ _" B! ?( |; h2 Z7 P  _$ X+------------+----------+-------+-------+
$ w" M" Y8 J1 Q
/ C6 `( G* k" z! x; @3 I+-------------+# A. o2 c) n$ k2 m- i6 T
| Standby MDS |
, s! ?3 f7 v$ S) \, t& _+-------------+
, m& ?5 M. D: j$ K+ ^9 n5 j( l|    ceph-2   |2 ~: p2 b: Z4 S3 i! V+ e
|    ceph-1   |  A0 N9 B8 h* c( @+ X, x
+-------------+
* P8 z% j" a$ ~6 L7 E9 ?* JMDS version: ceph version 12.2.8-52.el7 (3af3ca15b68572a357593c261f95038d02f46201) luminous (stable)
- Y6 I" Q9 Z5 f9 L* ?[root@ceph-3 test]# rados ls -p metedata |grep 2.00000000
% S- H" J, ]% A$ L. d9 t+ c, B3 j[root@ceph-3 test]# rados ls -p metedata |grep 2.00000000& j# [, [: V- K7 S
[root@ceph-3 test]# ls
6 ^5 \# X/ x* p: G4 w$ ]6 i9 f10000000000.00000000  recoverfile.png
$ c$ q0 ?- s& u1 @4 B) q[root@ceph-3 test]# , B7 U# v. J9 t/ i0 q
[root@ceph-3 test]# : \( ^# C8 p6 a& F2 v7 S7 ]
[root@ceph-3 test]# cd
# I+ b4 s; z" _$ ?[root@ceph-3 ~]# rados ls -p metedata
( Y& A1 U* v6 h( P6 ~7 ~601.00000000
5 ^1 _) }& t3 w600.00000000
2 r" c+ ^" Y, \' O* c1 m2 u603.00000000
& ]  Y) Q8 w3 Y: C% m1.00000000.inode4 p! ]7 Q, x" m0 g6 f
200.00000000
* X- i9 `: L1 ]% p$ K% ?" G200.00000001
7 `3 e0 \& o. @$ o5 A9 j606.00000000
( Y+ r5 @' N# m! l' k- H607.00000000
" Z2 {  n4 o6 c" F! e" E$ Y+ G' T2 a9 C608.00000000
6 r2 e3 J. a0 V) ^" c& E5 ^604.00000000* Z. w9 w1 a" ~
500.00000000$ k9 N- [  E! h" v+ J$ a3 B
mds_snaptable
- r7 E+ l+ j7 E605.00000000; M5 F8 A5 F" R8 P
mds0_inotable
+ a" v! O$ ~) ^- q* |  J100.00000000) O# q/ \# z# y( B
mds0_sessionmap: D; m' x! b5 n
609.00000000
3 }8 D& k$ J9 _! Y400.00000000% ]3 w+ Q: p, I4 H3 Q- ~
100.00000000.inode
  Z1 d, L" V; U/ O' t4 F1.00000000
: n9 q; `. h* i: V* t[root@ceph-3 ~]# rados ls -p metedata |grep 602
; I$ z" e; X. @! _) S[root@ceph-3 ~]# cd /mnt/^C
3 h1 ^7 [  L" ~' q8 H2 U[root@ceph-3 ~]# ceph -s
0 M8 O8 a* v6 x! C  cluster:8 B  T8 C. {9 A* \5 s$ G
    id:     57bf4711-2218-46af-99d6-9c68ae230ce1
( W- O: v+ }. k4 s' A    health: HEALTH_ERR4 f7 j9 N/ X! l8 n$ A, S
            1 filesystem is degraded: I. U5 x( h! ~# m
            1 mds daemon damaged3 D- x  I' E! i2 Y% N6 Y

# ?# j# j3 y6 Z' |  services:4 }/ H- g- }7 I4 \
    mon: 3 daemons, quorum ceph-2,ceph-1,ceph-38 T* y2 @: ?0 r
    mgr: ceph-2(active), standbys: ceph-1, ceph-3
& k: ?+ W) N# z/ W9 R- }2 P% ]2 T, W    mds: cephfs-0/1/1 up , 3 up:standby, 1 damaged$ j0 y  o; P- w/ F# Z0 O
    osd: 6 osds: 6 up, 6 in
# W: y- P- O/ x' B8 _
# s, V8 ]6 d' L1 R, n. Z4 c  data:
- l( |. ]; c& t" V9 T    pools:   2 pools, 256 pgs5 r& m# Z' @1 r. u$ Y( K
    objects: 21 objects, 50.4KiB! D+ \6 H8 g+ J( M8 [3 [
    usage:   738MiB used, 584GiB / 585GiB avail8 p/ Y9 _/ e& ?. E! z( _
    pgs:     256 active+clean9 p. T  ^% f- X6 P9 Q$ l+ y7 D
' {( x1 _6 f7 t$ X
[root@ceph-3 ~]# ls
% }2 k6 z" A& v2 J. }recovery_cephfs.py+ X' e  G8 L6 G) e$ Y9 b6 V# ]
[root@ceph-3 ~]# ls
& R: f- p' x  V- n' [recovery_cephfs.py& l" h, C9 R! I. A* p
[root@ceph-3 ~]# 4 G2 y3 m5 @7 u- \, b
[root@ceph-3 ~]# ls
7 q/ g1 L& R: o8 A/ Irecovery_cephfs.py, ?8 w4 a& ~# ?/ c
[root@ceph-3 ~]# df -TH
( H$ y4 D" [& w! |* T; L8 K4 a4 ?1 i) K; G1 c

3 v" q& h% U: W+ U6 ]  y; A# X, w
3 U3 X4 U6 `% Y( g) e^C9 P$ l0 T+ i# J0 I
[root@ceph-3 ~]# ls. m9 `) g# t7 S/ W& o- D
recovery_cephfs.py
8 ~! Y& @" ^3 E' z, e[root@ceph-3 ~]# ls
/ h/ g2 S% h6 I: Krecovery_cephfs.py
$ J9 Z% N; s, O8 b7 ]# \+ G[root@ceph-3 ~]# ls
* J: G1 u$ X) K/ M. ~. Y0 ?$ ]& jrecovery_cephfs.py  x) o/ R. t2 ~9 m
[root@ceph-3 ~]# ceph -s, d5 {' z/ }3 ^/ ?, j; M
  cluster:* D' l( |% s* S$ m2 b
    id:     57bf4711-2218-46af-99d6-9c68ae230ce1
# D; ~" F+ U2 u+ E: H% s: E8 I    health: HEALTH_ERR
' K! Q$ [* S: N" L3 `% n1 w            1 filesystem is degraded
% a; X9 I. u! D: h& n            1 mds daemon damaged
/ Y' {8 {: X+ F" i* [  [) D
# R0 s; o6 F1 @9 [1 @* s  services:
  |. `; T$ p+ ~, q$ L" @    mon: 3 daemons, quorum ceph-2,ceph-1,ceph-30 ?1 n1 u2 q; i& m+ f  F" L
    mgr: ceph-2(active), standbys: ceph-1, ceph-3
" E+ p8 y; I" a% m; O( b: b    mds: cephfs-0/1/1 up , 3 up:standby, 1 damaged
# b# ?  ?5 A8 c& y0 u  f    osd: 6 osds: 6 up, 6 in
+ [6 h# ~5 o$ k* j9 A9 U4 |' J' s/ V9 \/ v- D6 u8 f
  data:2 J5 ?( Q  Y+ ]8 z- A) A
    pools:   2 pools, 256 pgs
# n! n! i) A# H& d    objects: 21 objects, 50.4KiB% p7 L" z& J# ?" |
    usage:   738MiB used, 584GiB / 585GiB avail! k/ D7 u, m" R: b
    pgs:     256 active+clean
# n9 e. C9 y- g, p. N2 m/ E% x
+ Y. F* i9 z0 b9 |8 }: H8 ~! c[root@ceph-3 ~]# ceph fs status s7 e& E* W  ?$ @* W9 [4 y0 M

) x9 D6 P$ ^3 i% u5 ]3 d+-------------+# ?/ i' E/ ?* i. W: X6 }% o
| Standby MDS |( `7 c  F( i' z3 l6 y8 X) q
+-------------+8 W8 _5 k7 `; Z# y" D# `- e
|    ceph-2   |9 b! |' u2 u6 D
|    ceph-1   |
$ L% j5 ], z% b3 k  w/ e|    ceph-3   |2 O: P& {: ^5 g. d# y( A! c5 n
+-------------+
- R8 Z6 F& t* k' fMDS version: ceph version 12.2.8-52.el7 (3af3ca15b68572a357593c261f95038d02f46201) luminous (stable)
2 @0 Q! ]/ B/ t; x$ L# P[root@ceph-3 ~]# ceph fs status " p! _. w, u+ }, Q' `6 }+ f' w
cephfs - 0 clients6 e. S, e0 i. K3 V6 s) m8 v6 b. K
======" H6 g' M; j2 ]3 N0 m4 V# ~3 J9 p7 g
+------+--------+-----+----------+-----+------+
: f4 W$ P( i5 |! a| Rank | State  | MDS | Activity | dns | inos |
+ Q! T2 U2 n! K. C" B% U7 P# C& t+------+--------+-----+----------+-----+------+! W7 o( y9 W/ Q9 [0 |9 p
|  0   | failed |     |          |     |      |0 p- x3 v: L8 U3 N- ~4 `9 R
+------+--------+-----+----------+-----+------+' R0 y1 n) P- u9 W7 U! C
+------------+----------+-------+-------+7 }$ M& ^+ }" _" ]" s) u
|    Pool    |   type   |  used | avail |- T: i8 O# G0 @
+------------+----------+-------+-------+7 F( C5 Z: v3 g4 K  ]
|  metedata  | metadata | 6023  |  277G |
1 ]7 F1 C# J0 E7 a) U+ U9 F* z| cephfsdata |   data   | 44.5k |  277G |
* V! Z* N$ I2 D% J  i: _% z7 e( f5 E+------------+----------+-------+-------+7 c1 O7 q8 O4 b9 t9 j, J0 U

% L2 C( ?" r8 |9 ]5 E9 ^+-------------+) ?1 G, ]. n" A+ [4 @; u
| Standby MDS |
  w& |# H0 N! a+ B! n- R6 _+-------------+
4 ^" q5 p2 l" b2 W' _9 Y) i8 d, g|    ceph-2   |
8 ~) E" {/ ^/ @+ `|    ceph-1   |- s5 }$ u2 c5 J0 |6 X/ B  `
|    ceph-3   |
6 g- M2 r' {4 ~& a+-------------+% T) w3 e' z: o; \: R
MDS version: ceph version 12.2.8-52.el7 (3af3ca15b68572a357593c261f95038d02f46201) luminous (stable)4 K' o( A: O9 C- q, ]( j4 O% W9 V
[root@ceph-3 ~]# ceph fs status
0 a$ j3 v$ k3 U' A" h, q% Y  _9 [cephfs - 0 clients
1 y8 p  \  D2 V* k# j======$ a- d# x4 r$ A" a( W+ `' J
+------+--------+-----+----------+-----+------+
, r. w' b) k+ L4 y7 @% c| Rank | State  | MDS | Activity | dns | inos |& B$ k0 }1 R& Q  J) U2 {+ i
+------+--------+-----+----------+-----+------+
( q3 z! H4 g& H* S' Q9 ~6 Y|  0   | failed |     |          |     |      |
6 b( \4 d" k( L5 H# Q+------+--------+-----+----------+-----+------+
! t3 L5 F2 u, S" o( O" R. {. t8 R% r+------------+----------+-------+-------+
0 Y6 [" X' T( e5 T! Q|    Pool    |   type   |  used | avail |- q8 L" B0 w  |# c
+------------+----------+-------+-------+6 R2 ^# W  z& z/ T1 @$ J
|  metedata  | metadata | 6023  |  277G |8 p4 S( P7 q. I
| cephfsdata |   data   | 44.5k |  277G |9 ^( X& U9 J: t" q; l+ g' A/ b0 G6 ~
+------------+----------+-------+-------+" m0 b" D3 [! z2 D" W+ z; E! B
9 P; L8 F, s6 W+ y
+-------------+, M9 D* J% N! ?" M9 Y0 F7 }- Q3 U
| Standby MDS |% n6 ?5 V- F4 O, }
+-------------+) m: V; o! j0 ?2 s, t% ]2 `; }
|    ceph-2   |) z5 }, h. K' D. C( m
|    ceph-1   |" p* S& a: g4 P# g6 ?
|    ceph-3   |( N+ _' m# V: O1 D* o
+-------------+$ a/ q. M# m- B9 M+ g, q5 D
MDS version: ceph version 12.2.8-52.el7 (3af3ca15b68572a357593c261f95038d02f46201) luminous (stable)
. t( z0 E- l- D8 C[root@ceph-3 ~]# df -TH7 \' H$ K0 |1 S* o) [% N5 I5 ?1 x
^C: O9 l3 t$ N- m' V  {" m
[root@ceph-3 ~]# ls
# Z6 |7 F/ Y2 B- L) A6 ]recovery_cephfs.py8 m9 K, N) u# o- @
[root@ceph-3 ~]#
; i! B! `9 i8 J$ J
* A' ]% X( V* r- n- b2 q[root@ceph-3 ~]# ceph fs ls
. o3 v+ z0 ~$ g& Q& s6 L3 d- Yname: cephfs, metadata pool: metedata, data pools: [cephfsdata ]: O6 x- E& E# m# X
[root@ceph-3 ~]# ceph fs reset cephfs -* n' K5 I) b' w9 m3 N
--admin-daemon     -f                 -n                 -v                 --watch-info$ w3 n2 i$ ]! N+ i) P6 }) J
--admin-socket     --format           --name             --verbose          --watch-sec9 [( L2 [, P- u3 [
-c                 -h                 -o                 --version          --watch-warn
" F) S- D4 L" o8 _8 ]--cluster          --help             --out-file         -w                 ; w5 }6 M, H5 F) g
--concise          -i                 -s                 --watch            ) D& h4 o( r' @: w& Z0 K5 t
--conf             --id               --status           --watch-debug      
3 a$ o) ^% k: O5 l9 j1 h--connect-timeout  --in-file          --user             --watch-error      
# w. W( I  c2 i4 ~[root@ceph-3 ~]# ceph fs reset cephfs -
; m% l& x8 u- b$ I4 h--admin-daemon     -f                 -n                 -v                 --watch-info5 p& t# {" Y  w9 ^1 J6 F0 h
--admin-socket     --format           --name             --verbose          --watch-sec
6 Q7 ], R- p7 j  i4 p5 l-c                 -h                 -o                 --version          --watch-warn
  M; n( |- G+ \! v--cluster          --help             --out-file         -w                 ; I- d; [( Q/ o! t) ]
--concise          -i                 -s                 --watch            8 O6 K# v6 f' @" c
--conf             --id               --status           --watch-debug      
/ h' N* h: a' w! Q. o( X--connect-timeout  --in-file          --user             --watch-error      + h: U+ y: ^( L& X" Z8 q
[root@ceph-3 ~]# ceph fs reset cephfs --yes-i-really-mean-it 9 L0 i/ L8 O% t/ ]$ I8 t
                                                    #disaster recovery only: reset to a single-MDS map
( b: v, N' I! e0 T/ M[root@ceph-3 ~]# ceph fs reset cephfs --yes-i-really-mean-it . c$ a8 ?& K* W% ?' ]1 E$ g! _
[root@ceph-3 ~]# ceph fs ls5 Z5 a/ H) _/ Z. W; F: Y/ M  r
name: cephfs, metadata pool: metedata, data pools: [cephfsdata ]" o' {% F: P- p6 n9 J
[root@ceph-3 ~]# ceph fs status 0 f8 l9 ]4 g2 }6 ]9 w
cephfs - 1 clients
$ C/ u4 g( }+ e- m======/ C: H9 J% I& T0 c5 T
+------+--------+--------+----------+-------+-------+
( }6 C0 }$ Q7 F8 H8 x| Rank | State  |  MDS   | Activity |  dns  |  inos |5 s' d/ G& l1 R( t1 H7 j) @; d9 e
+------+--------+--------+----------+-------+-------+" A7 P9 T7 o: u' B! c/ c, I
|  0   | rejoin | ceph-3 |          |    1  |    3  |' t1 r5 c6 x! r  Y* ~
+------+--------+--------+----------+-------+-------+) v7 x' x& d9 d" F- A  h
+------------+----------+-------+-------+
; @! w& `7 u& S- G* M2 U|    Pool    |   type   |  used | avail |
' A0 |/ z( T  @4 h# Z+------------+----------+-------+-------+
1 ]0 `: H5 e' a  w|  metedata  | metadata | 6023  |  277G |9 F2 Z4 `& [2 H/ j
| cephfsdata |   data   | 44.5k |  277G |4 r+ \) Y' S4 O) ?6 C: j
+------------+----------+-------+-------+
0 ^6 H; C6 X/ }' l  G  G6 X- ^
6 g$ V* u5 r* `+-------------+
% o1 u1 q4 G0 i/ P% P- ^| Standby MDS |
, Y( t+ ]. v5 O: c+-------------+! Y) ]. z  t* s
|    ceph-2   |
7 u6 G- h: N5 X% U|    ceph-1   |
1 B2 \# q9 h" p( X; h' |6 D4 D+-------------+
( e8 ^/ p. S% S  M5 T3 ?MDS version: ceph version 12.2.8-52.el7 (3af3ca15b68572a357593c261f95038d02f46201) luminous (stable)
2 A0 \9 N" ~2 s[root@ceph-3 ~]# ceph fs status
/ a, `4 _! x& F* ~& H  i5 \) X2 Zcephfs - 1 clients) \/ w' U* V. b: G7 f
======
  v/ d0 d. j) z/ {7 g+------+--------+--------+---------------+-------+-------+* w6 d+ @: a0 \8 Z( Z/ _: s
| Rank | State  |  MDS   |    Activity   |  dns  |  inos |
/ X6 I( x5 E9 t2 h) Y$ j8 ~! V+------+--------+--------+---------------+-------+-------+1 X" h9 T7 a0 v3 F# p' P0 a/ M: X
|  0   | active | ceph-3 | Reqs:    0 /s |    1  |    3  |6 f2 `6 W( k  f# L& s
+------+--------+--------+---------------+-------+-------+
1 ~- U- S7 L. q/ F+------------+----------+-------+-------+
+ {2 @0 h- U; s9 [& E& @|    Pool    |   type   |  used | avail |; p; [' @6 U7 u$ u' g% m8 {4 s
+------------+----------+-------+-------+. T6 L! v/ q4 A5 G: v  u. Q
|  metedata  | metadata | 6135  |  277G |& b* c5 a# T- l2 ~5 O
| cephfsdata |   data   | 44.5k |  277G |
$ y6 {# b. [* O+------------+----------+-------+-------+
6 w0 ^% n7 D4 L* y, s
& S1 r3 M5 n) s4 f# ]+ w  ^6 a' A+-------------+
, Q; E9 _% `- ?1 n. x8 w| Standby MDS |
( C7 x: G. k9 E' v+-------------+
$ U+ N  V; P* q. @: [|    ceph-2   |
( X/ p# `9 h7 p. W- P0 j|    ceph-1   |
! d- a3 i2 m2 f( `) _+-------------+
: s  A( X* k/ ^5 x3 P2 qMDS version: ceph version 12.2.8-52.el7 (3af3ca15b68572a357593c261f95038d02f46201) luminous (stable)- p' [# r1 Q0 F: \7 a& @9 n
[root@ceph-3 ~]# 4 r, ]& P/ h% d& L6 h5 ?

1 v; E% K7 @3 H/ a8 _3 R
: n9 F3 y& I2 U  \

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:
  e3 n7 Q4 ~" {0 [* G7 I$ l8 u' F针对原因一:' B) ~% L8 g8 q9 t$ J0 s/ W
: ~& q2 R' u* Q1 i/ u$ @
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:1 X0 J% d  m/ p3 i$ {( a4 b
把 使用率比较高的osd降权重,手动触发osd的数据迁移。
4 n6 h2 P3 {4 Q, Q) [* t( ~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。% w: F) ?6 X& t" I
针对原因二(收效甚微):
. f5 V" i0 f0 w* x& Y3 R3 z: F: W- Y0 ~
调小mds mds_cache_memory_limit到40G 以下* k! L+ f1 F/ h& N
# ceph tell mds.\* injectargs '--mds_cache_memory_limit=40000000000'
! v0 M4 f# u8 |) x在/etc/ceph/ceph.conf修改,传到所有ceph节点2 F: P9 m6 O. Q( |, h
针对原因三(主要原因):
: M- a  u3 I2 G- L8 a; Q6 A9 ~
3 i! U4 b) P) J1 W* F7 L方法一 mds多活热备调低敏感度,有两个操作:
2 U( G+ W7 a6 M. r# N. r- a: e2 ~2 |: B5 X
调高mds_beacon_grace=300,调高超时时间。
$ W" h1 f6 T2 W- Y. Z
5 u! G+ Z& x2 d' z" `& z: W4 }/ s  写到/etc/ceph/ceph.conf里,传到所有mds节点,然后重启mds0 W. o: r1 Y6 o* o" f4 p4 f: p- O. L: s

; I+ \8 h, s/ l9 f: gmds采用热备模式,替换原来冷备模式,加快mds加入速度。
# e7 c1 [. Q0 ?% F3 F9 u& J5 h
% \4 v, y7 D/ n! s  ceph fs set <fs name> allow_standby_replay true
. C' B0 x5 y( X% X4 ~4 B% D/ M* G: o7 h! `
方法二 mds多活热备静态目录方法,即给目录划分不同的mds6 c& x9 ]9 f# F  k- |
9 \& d+ R( u' h6 p2 h6 G' D
命令:+ z) y# d4 N1 c
setfattr -n ceph.dir.pin -v 2 path/to/dir
2 K' @3 o' m$ ?  z$ Rmkdir -p a/b
3 C0 f1 R2 z0 R& u. ]. f- k  l( y# “a” and “a/b” both start without an export pin set0 |$ O9 p- x- e
setfattr -n ceph.dir.pin -v 1 a/. C4 R! }8 v* M+ s, A9 \' o8 @, c
# a and b are now pinned to rank 1
/ U" d" A& _! ]- I1 S1 ?: g! i0 N& O0 ysetfattr -n ceph.dir.pin -v 0 a/b
6 D$ F: V) E' \0 D% u: Z! p# n# a/b is now pinned to rank 0 and a/ and the rest of its children are still pinned to rank 1- \5 `' C' |8 o* Z* F
+ N* t# V8 Z5 m
" C1 A& k  U5 C0 c
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 23:02 , Processed in 0.044323 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