ceph集群警告和错误类型

指标	说明	级别
noscrub flag(s) set	防止集群做清洗操作
full flag(s) set	使集群到达设置的full_ratio值。会导致集群阻止写入操作
nodeep-scrub flag(s) set	防止集群进行深度清洗操作
pause flag(s) set	集群将会阻止读写操作，但不会影响集群的in、out、up或down状态。集群扔保持正常运行,就是客户端无法读写
noup flag(s) set	防止osd进入up状态
nodown flag(s) set	防止osd进入down状态
noout flag(s) set	防止osd进入out状态
noin flag(s) set	防止osd纳入ceph集群。有时候我们新加入OSD，并不想立马加入集群，可以设置该选项
nobackfill flag(s) set	防止集群进行数据回填操作
norebalance flag(s) set	防止数据均衡操作
norecover flag(s) set	避免关闭OSD的过程中发生数据迁移
notieragent flag(s) set
osds exist in the crush map but not in the osdmap	osd crush weight有值但是osd weight无值
application not enabled on 1 pool(s)	没有定义池的使用类型
osds have slow requests	慢查询
Monitor clock skew detected	时钟偏移
bigdata failing to advance its oldest client/flush tid	客户端和MDS服务器之间通信使用旧的tid
Many clients (34) failing to respond to cache pressure	如果某个客户端的响应时间超过了 mds_revoke_cap_timeout （默认为 60s ）这条消息就会出现
mons down, quorum	Ceph Monitor down
in osds are down	OSD down后会出现
cache pools are missing hit_sets	使用cache tier后会出现
has mon_osd_down_out_interval set to 0	has mon_osd_down_out_interval set to 0
is full	pool满后会出现
near full osd	near full osd
unscrubbed pgs	有些pg没有scrub
pgs stuck	PG处于一些不健康状态的时候，会显示出来
requests are blocked	slow requests会警告
osds have slow requests	slow requests会警告
recovery	需要recovery的时候会报
at/near target max	使用cache tier的时候会警告
too few PGs per OSD	每个OSD的PG数过少
too many PGs per OSD	too many PGs per OSD
> pgp_num	> pgp_num
has many more objects per pg than average (too few pgs?)	每个Pg上的objects数过多
no osds	部署完就可以看到，运行过程中不会出现
full osd	OSD满时出现
pgs are stuck inactive for more than	Pg处于inactive状态，该Pg读写都不行
scrub errors	scrub 错误出现，是scrub错误?还是scrub出了不一致的pg