|
|
[root@ceph2 ceph]# ceph -s
* L. B1 j! N, Y |) O/ B cluster:1 u+ a- d& w% H# F
id: 2f6bb77f-8b7e-4530-bd4b-8e6f6ab58833, O2 b# b. s" m9 p5 \5 [" o
health: HEALTH_WARN& G9 w9 r( \* L+ Q
all OSDs are running pacific or later but require_osd_release < pacific
: z/ [6 Q" e U' O3 ^# w
( S+ F3 f2 p1 v* X* o6 k& z: h services:
1 f1 x% j2 P' v2 S; ~7 z mon: 3 daemons, quorum ceph1,ceph2,ceph3 (age 42m)" n% r) P& Q4 c# Q
mgr: ceph3(active, since 12m), standbys: ceph2, ceph1- D* W( a& y( g& E8 L
mds: 1/1 daemons up, 2 standby
7 c* Q# K: b- |9 M1 `( B osd: 9 osds: 9 up (since 36m), 9 in (since 10h)% d& {2 W& x: D, S* t2 }# w! y9 t
rgw: 3 daemons active (3 hosts, 1 zones)
2 x8 E& h+ ? r4 u/ \- V4 _. H% n6 k
data:
5 C$ `) h3 c' ? volumes: 1/1 healthy; e8 ^& @ d' e8 m/ U$ D
pools: 9 pools, 289 pgs7 Z. l* D) M: \
objects: 735 objects, 2.0 GiB$ s2 C+ [/ Q u/ R
usage: 51 GiB used, 174 GiB / 225 GiB avail" n0 }0 e4 v& Y% w
pgs: 289 active+clean+ j' k( U& o% X; V
" \% B7 s" x( x! `# U% m: R
直译所有 OSD 实际运行的版本已经是 Pacific 或更高
- M, B$ I( c q1 V6 }. ?0 }但集群配置里仍然要求 require_osd_release < pacific
人话OSD 已经升级完了，但集群“版本锁”还停留在老版本
Ceph 为了防止误升级导致数据不可逆，升级是两阶段的：
( Q. o! ?% i$ F+ ^! }) Y6 I; r G* {% D先升级 daemon（你已经完成） 再手动提升集群最小版本要求（你还没做） 6 }: c! b/ }4 x. Q x( ^
: K; J: t4 G5 W- M( C" O
8 {1 K+ T. w6 K* G[root@ceph2 ceph]# ceph osd require-osd-release pacific# N% x) e% f K; ~% P1 T# e
! M; L6 U- e3 s) G# ?* [
再次查看状态：0 E# N6 K$ m2 S
[root@ceph2 ceph]# ceph -s0 w$ W% [- H( e0 ?! P+ r
cluster:# T2 Y" t6 G, F. F7 P# f+ y& _
id: 2f6bb77f-8b7e-4530-bd4b-8e6f6ab58833+ \4 n7 y$ z* ?( p. V& G' t
health: HEALTH_OK
) d9 J3 [3 ]+ q2 c# |! B# A7 P( e1 V7 m. E
services:
. d$ a5 a9 M" }, @ mon: 3 daemons, quorum ceph1,ceph2,ceph3 (age 43m)
6 e/ n1 n0 m* C1 z. q* n7 j mgr: ceph3(active, since 13m), standbys: ceph2, ceph1
' ?7 o% T9 h" v1 s* I. V4 B$ Z mds: 1/1 daemons up, 2 standby C9 d+ i2 H% U3 y) K4 l& T/ R
osd: 9 osds: 9 up (since 37m), 9 in (since 10h)/ S; r2 O. m! t+ X: L( `
rgw: 3 daemons active (3 hosts, 1 zones)9 d: _- W" `1 X8 @
6 _' _6 R' T$ k( W+ N
data:
# q, r9 o- S7 N1 Y3 f7 f volumes: 1/1 healthy
+ z1 N! p9 y8 `9 e6 d pools: 9 pools, 289 pgs2 K$ T& D2 I+ o8 s' A6 E: X
objects: 735 objects, 2.0 GiB6 a, c! v( ^/ }" }8 g
usage: 51 GiB used, 174 GiB / 225 GiB avail& N' k8 b& ^+ F& j3 V; {
pgs: 289 active+clean* X2 w) e, w/ w9 \, s" j
3 s; \4 G5 S; g" b$ ?% o+ r: `
# _4 {" v2 e$ i 这是一次性、不可回退操作: S% b* N3 q& I: F, P" n
但你的前提已经满足（全部 OSD 都是 pacific+），所以是安全的。
8 B; P6 s) \; D8 D$ ~9 |7 J( T: L
8 r% R z0 h7 O) }( i+ R: \' A
# ceph osd versions
V. }% I0 j# |# E$ V# D) ?{
) c7 h* ?2 D( @# E% V+ l+ t1 ? "ceph version 16.2.10-275.el8cp (69b25b314048bb8542dce8512a249fc34a7add1d) pacific (stable)": 90 w: s' C7 R5 E/ w
}
8 w3 J. V0 L) d8 ?4 _ m7 c5 R6 \: B, C4 F9 @: d+ P0 f
/ K9 t. _- H B) b" e
: _& ?5 V. B6 w! d* r3 v* q& g( @- Q7 G
9 o9 K+ C" L/ Z# e. e. U o |
|
发表于 2025-12-30 00:15:56