ceph：370 pgs 卡在不活动状态超过 300 秒答案

【问题标题】：ceph: 370 pgs are stuck inactive for more than 300 secondsceph：370 pgs 卡在不活动状态超过 300 秒
【发布时间】：2017-06-29 08:02:58
【问题描述】：

我有一个小型 Ceph 集群。这里描述了它的设置方式：

https://www.theo-andreou.org/?p=1750

在部署节点（托管 ntp 服务器的位置）上重新启动后，我得到：

ceph health; ceph osd tree
HEALTH_ERR 370 pgs are stuck inactive for more than 300 seconds; 370 pgs stale; 370 pgs stuck stale; too many PGs per OSD (307 > max 300)
ID  WEIGHT   TYPE NAME       UP/DOWN REWEIGHT PRIMARY-AFFINITY
 -1 10.88989 root default                                      
 -2  0.54449     host node02                                  
  0  0.54449         osd.0      down        0          1.00000
 -3  0.54449     host node03                                  
  1  0.54449         osd.1      down        0          1.00000
 -4  0.54449     host node04                                  
  2  0.54449         osd.2      down        0          1.00000
 -5  0.54449     host node05                                  
  3  0.54449         osd.3      down        0          1.00000
 -6  0.54449     host node06                                  
  4  0.54449         osd.4      down        0          1.00000
 -7  0.54449     host node07                                  
  5  0.54449         osd.5      down        0          1.00000
 -8  0.54449     host node08                                  
  6  0.54449         osd.6      down        0          1.00000
 -9  0.54449     host node09                                  
  7  0.54449         osd.7      down        0          1.00000
-10  0.54449     host node10                                  
  8  0.54449         osd.8      down        0          1.00000
-11  0.54449     host node12                                  
  9  0.54449         osd.9      down        0          1.00000
-12  0.54449     host node13                                  
 10  0.54449         osd.10     down        0          1.00000
-13  0.54449     host node14                                  
 11  0.54449         osd.11     down        0          1.00000
-14  0.54449     host node16                                  
 12  0.54449         osd.12     down        0          1.00000
-15  0.54449     host node17                                  
 13  0.54449         osd.13     down        0          1.00000
-16  0.54449     host node18                                  
 14  0.54449         osd.14     down        0          1.00000
-17  0.54449     host node19                                  
 15  0.54449         osd.15       up  1.00000          1.00000
-18  0.54449     host node20                                  
 16  0.54449         osd.16       up  1.00000          1.00000
-19  0.54449     host node21                                  
 17  0.54449         osd.17       up  1.00000          1.00000
-20  0.54449     host node22                                  
 18  0.54449         osd.18       up  1.00000          1.00000
-21  0.54449     host node23                                  
 19  0.54449         osd.19       up  1.00000          1.00000

节点已启动并可通过 ssh 访问。有没有办法让系统恢复健康？

【问题讨论】：

标签： cluster-computing ceph

【解决方案1】：

显然 OSD 守护程序已关闭（即使在报告为“启动”的节点上也是如此）。运行 I=0; for ID in {02..10} {12..14} {16..23}; do ceph-deploy osd activate node${ID}:/var/local/osd${I}; I=$((${I}+1)); done 后，我现在有了 HEALTH_OK

非常感谢#ceph IRC 频道！

【讨论】：