openstack集成ceph
这里不记录ceph安装的过程了，ceph安装详情见这里：http://www.vpsee.com/2015/07/install-ceph-on-centos-7/
ceph官方安装文档：http://docs.ceph.com/ceph-deploy/docs/install.html
ceph常用命令：http://zhanguo1110.blog.51cto.com/5750817/1543032
安装ceph client
# 创建一个pool（具体根据你实际pg来，使用ceph-deploy安装完的时候会自动帮你创建一个rbd的pool） # 这里glance、cinder、nova共用一个pool，实际生产环境 [root@ceph01 ~(keystone_admin)]# ceph osd pool create rbd 128 # glance-api、nova-compute、cinder-backup、cinder-volume节点安装ceph client包 [root@ceph01 ~(keystone_admin)]# yum install python-rbd ceph
建立ceph client认证
# 创建ceph认证的用户，这里glance、cinder、nova共用一个ceph认证用户
# 官方文档上建议分别为nova、cinder、glance创建不同的用户
[root@ceph01 ~(keystone_admin)]# ceph auth get-or-create client.rbd mon 'allow r' osd 'allow class-read object_prefix rbd_children, allow rwx pool=rbd'
# 查看生成的rbd用户keyring
[root@ceph01 ~(keystone_admin)]# ceph auth get-or-create client.rbd
[client.rbd]
key = AQBKGHBWzJCYORAAABHki+tWoOFgiTZL8FNnaA==
# 以下两步操作，在glance-api、cinder-volume、cinder-backup、nova-compute节点上执行
# 创建keyring文件，添加如下内容
[root@ceph01 ~(keystone_admin)]# vim /etc/ceph/ceph.client.rbd.keyring
[client.rbd]
key = AQBKGHBWzJCYORAAABHki+tWoOFgiTZL8FNnaA==
# 因为nova、cinder、glance共用一个用户，所以文件权限改为777
[root@ceph01 ~(keystone_admin)]# ll /etc/ceph/ceph.client.rbd.keyring
-rwxrwxrwx 1 root root 61 Dec 15 21:44 /etc/ceph/ceph.client.rbd.keyring
# 配置libvirt secret key，libvirt进程需要cinder keyring(这里也就是client.rbd)
# 这样它才能访问ceph集群挂载块设备
# 使用tee命令创建一个暂时的文件
[root@ceph01 ~(keystone_admin)]# ceph auth get-key client.rbd | ssh {your-compute-node} tee client.cinder.key # 针对所有计算节点
# 以下操作在所有计算节点上执行
[root@ceph01 ~(keystone_admin)]# uuidgen # 生成随机的uuid
aa03e7e8-6fcc-443f-94aa-ac169bfd0fd5
cat > secret.xml <<EOF
<secret ephemeral='no' private='no'>
<uuid>aa03e7e8-6fcc-443f-94aa-ac169bfd0fd5</uuid>
<usage type='ceph'>
<name>client.rbd secret</name>
</usage>
</secret>
EOF
sudo virsh secret-define --file secret.xml
Secret aa03e7e8-6fcc-443f-94aa-ac169bfd0fd5 created
sudo virsh secret-set-value --secret aa03e7e8-6fcc-443f-94aa-ac169bfd0fd5 --base64 $(cat client.rbd.key) && rm client.rbd.key secret.xml
# 实际上计算节点的uuid可以不一致，保持一致单纯只是从平台一致性来考虑的。
OpenStack rbd配置
# glance rbd配置
[root@ceph01 ~(keystone_admin)]# vim /etc/glance/glance-api.conf
[DEFAULT]
show_image_direct_url = True # 启动镜像copy-on-write克隆功能
[glance_store]
default_store = rbd
stores = rbd
filesystem_store_datadir=/var/lib/glance/images/
rbd_store_pool = rbd
rbd_store_user = rbd
rbd_store_ceph_conf = /etc/ceph/ceph.conf
rbd_store_chunk_size = 8
[paste_deploy]
flavor = keystone # 禁用glance cache管理，如果你的flavor=keystone+cachemanagement,请修改
# cinder rbd配置
[root@ceph01 ~(keystone_admin)]# vim /etc/cinder/cinder.conf
volume_driver = cinder.volume.drivers.rbd.RBDDriver
rbd_pool = rbd
rbd_ceph_conf = /etc/ceph/ceph.conf
rbd_flatten_volume_from_snapshot = false
rbd_max_clone_depth = 5
rbd_store_chunk_size = 4
rados_connect_timeout = -1
glance_api_version = 2 # 官方文档说如果你配置了cinder multi backends，必须配置这个
rbd_user = rbd
rbd_secret_uuid = aa03e7e8-6fcc-443f-94aa-ac169bfd0fd5
# nova rbd配置
[libvirt]
inject_password = False # openstack boot from volume启动instance的时候不支持file injection
inject_key = False # ditto
inject_partition = -2 # ditto
virt_type = kvm
live_migration_flag=VIR_MIGRATE_UNDEFINE_SOURCE,VIR_MIGRATE_PEER2PEER,VIR_MIGRATE_LIVE,VIR_MIGRATE_PERSIST_DEST
images_type = rbd
images_rbd_pool = rbd
images_rbd_ceph_conf = /etc/ceph/ceph.conf
rbd_user = rbd
rbd_secret_uuid = aa03e7e8-6fcc-443f-94aa-ac169bfd0fd5
disk_cachemodes="network=writeback"
# 每个计算节点上执行，编辑ceph配置文件
[root@ceph01 ~(keystone_admin)]# vim /etc/ceph/ceph.conf # 开启admin socket，有助于排错
[client]
rbd cache = true
rbd cache writethrough until flush = true
admin socket = /var/run/ceph/guests/$cluster-$type.$id.$pid.$cctid.asok
log file = /var/log/qemu/qemu-guest-$pid.log
rbd concurrent management ops = 20
[root@ceph01 ~(keystone_admin)]# mkdir -p /var/run/ceph/guests/ /var/log/qemu/
[root@ceph01 ~(keystone_admin)]# chown qemu:qemu /var/run/ceph/guests /var/log/qemu/
OpenStack 配置最佳实践（转载自：http://www.wzxue.com/openstack-ceph-kilo/）
Ceph.conf :
[client]
rbd cache = true
rbd cache writethrough until flush = true
rbd concurrent management ops = 20
admin socket = /var/run/ceph/$cluster-$type.$id.$pid.$cctid.asok
log file = {{ rbd_client_log_file }}
GLANCE
Disable local cache: s/flavor = keystone+cachemanagement/flavor = keystone/
Expose images URL: show_image_direct_url = T
w_scsi_model=virtio-scsi # for discard and perf
hw_disk_bus=scsi
Nova:
hw_disk_discard = unmap # enable discard support (be careful of perf)
inject_password = false # disable password injection
inject_key = false # disable key injection
inject_partition = -2 # disable partition injection
disk_cachemodes = "network=writeback" # make QEMU aware so caching works
live_migration_flag="VIR_MIGRATE_UNDEFINE_SOURCE,VIR_MIGRATE_PEER2PEER,
VIR_MIGRATE_LIVE,VIR_MIGRATE_PERSIST_DEST"
Cinder:
glance_api_version = 2
# 最后重启下服务
[root@ceph01 ~(keystone_admin)]# service openstack-glance-api restart
[root@ceph01 ~(keystone_admin)]# service openstack-nova-compute restart
[root@ceph01 ~(keystone_admin)]# service openstack-cinder-volume restart
参考链接
http://docs.ceph.com/docs/master/rbd/rbd-openstack/
http://my.oschina.net/JerryBaby/blog/376580?fromerr=wNPJrqPP#OSC_h2_1