ceph设计原理与实现(RGW、CephFS、实战)

1、侠义讲，对象存储就是云存储 2、分布式文件系统：CephFS 3、每天晚上10点到早上7点Scrub ceph tell osd.* injectargs "--osd-scrub-begin-hour 22" ceph tell osd.* injectargs "--osd-scrub-end-hour 7" vi /etc/ceph/ceph.conf [osd] osd_scrub_begin_hour = 22 osd_scrub_end_hour = 7 4、ceph数据重建配置策略 //提升recovery的优先级，降低client的优先级 ceph tell osd.* injectargs "--osd-recovery-op-priority 63" ceph tell osd.* injectargs "--osd-client-op-priority 3" //等recovery完成， ceph tell osd.* injectargs "--osd-recovery-op-priority 3" ceph tell osd.* injectargs "--osd-client-op-priority 63" //永久性修改优先级 /etc/ceph/ceph.conf

image.png

5、禁止数据重建和迁移、开启数据重建和迁移 ceph osd set norebalance ceph osd set norecovery ceph osd set nobackfill ////启用 ceph osd unset norebalance ceph osd unset norecovery ceph osd unset nobackfill 6、查看、管理full和near_full告警 查询：ceph --admin-daemon /run/ceph/ceph-mon.yourmonip.asok config show|grep full_ratio ceph --admin-daemon /run/ceph/ceph-osd.yourosdip.asok config show|grep full //遇见full，后处理 ceph osd pause ceph tell mon.* injectargs "--mon-osd-full-ratio 0.96" ceph tell osd.* injectargs "--mon-osd-full-ratio 0.96" ceph pg set_full_ratio 0.96 ceph osd unpause 7、增量备份 rdb export rdb/test_img Time1_Data_img

image.png

image.png

image.png

8、查找image被多少客户端连接，将这些客户端加入黑名单

rdb status test_img

rdb info test_img

rdb image test_img

rados listwatchers -p rdb rdb_header.xxxx watcher=xx client.xxx cookie=xxx

ceph osd blacklist add xxx blacklisting xx until xx

9、ceph对接openstack

image.png

image.png

image.png