English

Openstack Cheat Sheet

OpenStack项目

当前最新为Ocata版本, 数据来源于https://www.openstack.org/software/project-navigator, 点击查看完整JSON数据。

管理工具

名称	LOGO	服务	年龄	部署率	成熟度
Horizon		Dashboard	6年	87%	6/7
OSclient (CLI)		Command-line client	1年	0%	0/7
Rally		Benchmark service	1年	23%	0/7
Senlin		Clustering service	1年	0%	0/7
Vitrage		RCA (Root Cause Analysis service)	1年	0%	0/7
Watcher		Optimization Service	1年	0%	0/7

网络 & 内容分发

名称	LOGO	服务	年龄	部署率	成熟度
Neutron		Networking	5年	93%	7/7
Designate		DNS Service	3年	16%	3/7
Dragonflow		Neutron Plugin	2年	0%	0/7
Kuryr		Container plugin	1年	0%	1/7
Octavia		Load Balancer	1年	0%	1/7
Tacker		NFV Orchestration	1年	0%	1/7
Tricircle		Networking Automation for Multi-Region Deployments	1年	0%	0/7

部署工具

名称	LOGO	服务	年龄	部署率	成熟度
Chef OpenStack		Chef cookbooks for OpenStack	1年	0%	1/7
Kolla		Container deployment	1年	12%	1/7
OpenStack Charms		Juju Charms for OpenStack	1年	0%	0/7
OpenStack-Ansible		Ansible Playbooks for OpenStack	1年	0%	0/7
Puppet OpenStack		Puppet Modules for OpenStack	1年	0%	0/7
TripleO		Deployment service	1年	9%	0/7

安全 & 认证

名称	LOGO	服务	年龄	部署率	成熟度
Keystone		Identity service	6年	96%	6/7
Barbican		Key Management	4年	9%	4/7
Congress		Governance	2年	2%	1/7
Mistral		Workflow service	1年	5%	1/7

数据 & 分析

名称	LOGO	服务	年龄	部署率	成熟度
Trove		Database as a Service	4年	13%	3/7
Sahara		Big Data Processing Framework Provisioning	3年	10%	3/7
Searchlight		Indexing and Search	1年	0%	0/7

存储 & 备份 & 恢复

名称	LOGO	服务	年龄	部署率	成熟度
Swift		Object Store	7年	52%	6/7
Cinder		Block Storage	5年	88%	7/7
Manila		Shared Filesystems	3年	14%	5/7
Karbor		Application Data Protection as a Service	1年	0%	0/7
Freezer		Backup, Restore, and Disaster Recovery	2年	0%	0/7

计算

名称	LOGO	服务	年龄	部署率	成熟度
Nova		Compute Service	7年	95%	7/7
Glance		Image Service	7年	95%	5/7
Ironic		Bare Metal Provisioning Service	3年	21%	5/7
Magnum		Container Orchestration Engine Provisioning	2年	11%	2/7
Storlets		Computable Object Store	1年	0%	0/7
Zun		Container Management Service	1年	0%	1/7

应用服务

名称	LOGO	服务	年龄	部署率	成熟度
Heat		Orchestration	5年	67%	5/7
Zaqar		Messaging Service	3年	4%	4/7
Murano		Application Catalog	2年	11%	1/7
Solum		Software Development Lifecycle Automation	1年	2%	0/7

监控 & 计量

名称	LOGO	服务	年龄	部署率	成熟度
Ceilometer		Metering & Data Collection Service	5年	55%	1/7
CloudKitty		Billing and chargebacks	1年	0%	0/7
Monasca		Monitoring	1年	0%	0/7
aodh		Alarming Service	1年	0%	0/7
panko		Event, Metadata Indexing Service	7年	0%	0/7

OpenStack服务与AWS服务对比

官方资源

• 主页
• git源码地址
• Github源码镜像
• 组件项目 (Big Tent Governance)
• 官方文档
• 邮件列表
• IRC(freenode)
• Code review
• Roadmap
• Wiki
• Blueprints
• Bugs
• 支持公司 (Companies Supporting The OpenStack Foundation)
• 贡献统计(来源于stacklytics) (Provide transparent and meaningful statistics regarding contributions to both OpenStack itself and projects related to OpenStack.)
• 社区开发文档(一步步教你如何加入社区开发)
• 代码提交测试(新手代码提交测试，新手可以通过此熟悉代码提交流程)
• git commit message规范(提交代码时必须遵循的message的规范)
• Openstack代码规范(Openstack代码规范，开发人员必读)
• Openstack各个组件的成熟度和热度统计
• Openstack版本列表(可以查看数字版本对应Openstack版本代号，如Nova版本的13.x.x对应mitaka版本)

成员

白金会员

Provide a significant portion of the funding to achieve the Foundation's mission of protecting, empowering and promoting the OpenStack community and software. Each Platinum Member's company strategy aligns with the OpenStack mission and is responsible for committing full-time resources toward the project. There are eight Platinum Members at any given time, each of which holds a seat on the Board of Directors.

• AT&T
• Canonical
• Hewlett Packard Enterprise
• IBM
• Intel
• Rackspace
• Red Hat, Inc.
• SUSE

黄金会员

Provide funding and pledge strategic alignment to the OpenStack mission. There can be up to twenty-four Gold Members at any given time, subject to board approval.

• Aptira
• CCAT
• Cisco
• Dell
• DreamHost
• EasyStack
• EMC
• Ericsson
• Fujitsu
• Hitachi
• Huawei
• inwinSTACK
• Juniper Networks
• Mirantis
• NEC
• NetApp
• Symantec
• UnitedStack Inc.
• Virtuozzo
• Yahoo Inc.

自动化部署

• Devstack(Oneiric powered development environment for openstack.)
• Fuel(The leading purpose-built open source deployment and management tool for OpenStack.)
• RDO(A communiaty of people using and deploying OpenStack on CentOS, Fedora, and Red Hat Enterprise Linux.)
• openstack-puppet(Bring scalable and reliable IT automation to OpenStack cloud deployments using puppet.)
• openstack-ansible(Ansible playbooks for deploying OpenStack.)
• kolla(Deploying OpenStack using Docker.)
• TripleO(A program aimed at installing, upgrading and operating OpenStack clouds using OpenStacks own cloud facilities as the foundations.)
• SaltStack(Deploying Openstack using saltstack.)
• packstack (Install utility to deploy openstack on multiple hosts.)

监控

• cloudpulse: Openstack服务状态监控
• Vitrage: Openstack故障分析, 故障生成树
• fuel-plugin-external-zabbix: Fuel Openstack zabbix监控项目,包含所有的监控项
• fuel-ostf: Fuel的集成测试、功能测试框架

竞争对手

• AWS(Offers reliable, scalable, and inexpensive cloud computing services.)
• CloudStack(Open source software designed to deploy and manage large networks of virtual machines, as highly scalable Infrastructure as a Service (IaaS) cloud computing platform.)
• Vmware(The industry-leading virtualization software company.)
• Docker(Docker containers and tooling make building and shipping applications dramatically easier and faster.)

服务目录

服务初始化模板--以cinder为例

初始化数据库:

CREATE DATABASE cinder;
GRANT ALL PRIVILEGES ON cinder.* TO 'cinder'@'localhost' IDENTIFIED BY 'CINDER_DBPASS';
GRANT ALL PRIVILEGES ON cinder.* TO 'cinder'@'%' IDENTIFIED BY 'CINDER_DBPASS';

创建账户:

openstack user create --domain default --password CINDER_PASS cinder
openstack role add --project service --user cinder admin

创建endpoints:

openstack service create --name cinder \
  --description "OpenStack Block Storage" volume

openstack endpoint create --region RegionOne \
  volume public 'http://controller:8776/v2/%(tenant_id)s'

openstack endpoint create --region RegionOne \
  volume admin 'http://controller:8776/v2/%(tenant_id)s'

openstack endpoint create --region RegionOne \
  volume internal 'http://controller:8776/v2/%(tenant_id)s'

keystore

glance

• OpenStack Virtual Machine Image Guide: OpenStack镜像制作指南，包括OpenStack镜像必须满足的要求、修改镜像的方法（挂载raw、qcow2格式镜像文件）。

nova

创建Ceph secret:

#!/bin/sh

USERNAME=${1:-admin}
UUID=${2:-$(uuidgen)}

# 生成secret.xml文件
cat > secret.xml <<EOF
<secret ephemeral='no' private='no'>
  <uuid>${UUID}</uuid>
  <usage type='ceph'>
    <name>client.${USERNAME} secret</name>
  </usage>
</secret>
EOF

# 基于xml文件创建secret
UUID=$(sudo virsh secret-define --file secret.xml | grep -Po '\w{8}-(\w{4}-){3}\w{12}')
if [ $? -ne 0 ]; then
    rm -f client."${USERNAME}".key secret.xml
    exit 1
fi

# 获取ceph client的密钥
KEY=$(ceph auth get-key client."${USERNAME}")
if [ $? -ne 0 ]; then
    sudo virsh secret-undefine "${UUID}"
    exit 1
fi

# 关联密钥到新创建的secret
sudo virsh secret-set-value --secret "${UUID}" --base64 "${KEY}" >/dev/null \
    && rm -f client."${USERNAME}".key secret.xml

echo "Secret $UUID create sucessfully!"

👉scripts/create_libvirt_ceph_secret.sh.

Tips

列出所有的错误云主机:

nova list --status error --all-tenants

列出主机和映射的宿主机:

nova list --fields name,host

获取所有主机的id:

nova list --minimal | tail -n +4 | head -n -1 | cut -d ' ' -f 2

获取所有error状态的主机id:

nova list --minimal --status error --all-tenants | tail -n +4 | head -n -1 | cut -d ' ' -f 2

获取所有名称以"test"开头的主机:

nova list --name "^test"

获取所有主机挂载的volume id:

nova show --minimal 65e761b3-63b4-498f-9735-08e246c1e976 | grep 'os-extended-volumes:volumes_attached' | cut -d '|' -f 3 | jq '.[].id' | sed 's/"//g'

获取所有主机挂载的volume id以及设备名称:

nova show --minimal | grep 'os-extended-volumes:volumes_attached' | cut -d '|' -f 3 | jq '.[]|.id + " " + .device_name' | sed 's/"//g'

Cinder

CinderSupportMatrix:

backend配置

LVM

[lvm]
volume_backend_name=lvm
volume_driver=cinder.volume.drivers.lvm.LVMVolumeDriver
iscsi_helper=lioadm
volume_group=cinder-volumes
volumes_dir=/var/lib/cinder/volumes

RBD

[ceph]
volume_backend_name = ceph
volume_driver = cinder.volume.drivers.rbd.RBDDriver
rbd_ceph_conf = /etc/ceph/ceph.conf
rbd_pool = cinder_pool
rbd_user = admin
rbd_secret_uuid = e3e0e2ea-a8ca-4f13-9528-5d5d2d813bc2

创建volume-type:

cinder type-create lvm
cinder type-key lvm set volume_backend_name=lvm
cinder type-create ceph
cinder type-key ceph set volume_backend_name=ceph

挂载volume到本地

安装cinderclient扩展包:

pip install python-brick-cinderclient-ext

attach volume到本地:

$ cinder local-attach 9770a53e-91ce-4cd0-afde-5793bfa6b0ef

+----------+-----------------------------------+
| Property | Value                             |
+----------+-----------------------------------+
| path     | /dev/sdb                          |
| scsi_wwn | 360000000000000000e00000000010001 |
| type     | block                             |
+----------+-----------------------------------+

挂载volume:

$ mkfs.ext4 /dev/sdb
mke2fs 1.42.13 (17-May-2015)
Creating filesystem with 262144 4k blocks and 65536 inodes
Filesystem UUID: d53932de-2844-4f63-8b37-67db52d1a243
Superblock backups stored on blocks:
        32768, 98304, 163840, 229376

Allocating group tables: done
Writing inode tables: done
Creating journal (8192 blocks): done
Writing superblocks and filesystem accounting information: done

$ mount /dev/sdb /mnt

umount & detach:

umount /mnt
cinder local-detach 9770a53e-91ce-4cd0-afde-5793bfa6b0ef

neutron

swift

trove

sahara

镜像列表

• sahara-image-elements(Disk image elements for Sahara.)
• 如果不想自己做镜像，这里提供现成的镜像下载: prepared images.
• Building Images for Vanilla Plugin.
• Building Images for Cloudera Plugin.

创建集群--以spark 1.6.0为例

初始化变量:

# 集群名字
NAME=test
# 镜像uuid，必须是通过sahara注册过的镜像
IMAGE_ID=6ba27a59-ecd7-47d4-8df4-925acb23bb87
# 浮动IP网络uuid
FLOATING_IP=7517cbd2-2d07-44ad-8eaa-b9ec7098071a
# flavor id
FLAVOR_ID=2
# 密钥名称
KEYPAIR_NAME=server-230
# 租户网络uuid
NETWORK_ID=e478bc87-a067-402f-98e2-9bdc9b180d5e

创建master group:

# Create master node group template
cat > spark-master-node-group-template.json <<EOF
{
    "auto_security_group": true,
    "availability_zone": "",
    "description": "spark 1.6.0 master node group",
    "flavor_id": "${FLAVOR_ID}",
    "floating_ip_pool": "${FLOATING_IP}",
    "hadoop_version": "1.6.0",
    "image_id": "${IMAGE_ID}",
    "is_protected": false,
    "is_proxy_gateway": false,
    "is_public": true,
    "name": "${NAME}-master",
    "node_configs": {
        "HDFS": {}
    },
    "node_processes": [
        "namenode",
        "master"
    ],
    "plugin_name": "spark",
    "security_groups": [],
    "shares": [],
    "use_autoconfig": true,
    "volumes_per_node": 1,
    "volumes_size": 20
}
EOF
cat spark-master-node-group-template.json
sahara node-group-template-create --json spark-master-node-group-template.json 2>/dev/null

创建slave group:

# Create worker node group template
cat > spark-worker-node-group-template.json <<EOF
{
    "auto_security_group": true,
    "availability_zone": "",
    "description": "spark 1.6.0 master node group",
    "flavor_id": "2",
    "floating_ip_pool": "${FLOATING_IP}",
    "hadoop_version": "1.6.0",
    "image_id": "${IMAGE_ID}",
    "is_protected": false,
    "is_proxy_gateway": false,
    "is_public": true,
    "name": "${NAME}-worker",
    "node_configs": {
        "HDFS": {}
    },
    "node_processes": [
        "datanode",
        "slave"
    ],
    "plugin_name": "spark",
    "security_groups": [],
    "shares": [],
    "use_autoconfig": true,
    "volumes_per_node": 1,
    "volumes_size": 20
}
EOF
cat spark-worker-node-group-template.json
sahara node-group-template-create --json spark-worker-node-group-template.json

创建集群模板:

SPARK_MASTER_GROUP_ID=$(sahara node-group-template-list  2>/dev/null | grep -P -- "\s$NAME-master\s" | awk '{print $4}')
SPARK_WORKER_GROUP_ID=$(sahara node-group-template-list  2>/dev/null | grep -P -- "\s$NAME-worker\s" | awk '{print $4}')

# Create cluster template
cat >spark-cluster-template.json <<EOF
{
    "anti_affinity": [],
    "cluster_configs": {
        "HDFS": {
            "dfs.replication": 1
        },
        "Spark": {},
        "general": {
            "Enable XFS": true
        }
    },
    "description": "spark 1.6.0",
    "hadoop_version": "1.6.0",
    "is_protected": false,
    "is_public": true,
    "name": "${NAME}-cluster-template",
    "node_groups": [
        {
            "count": 1,
            "name": "${NAME}-master",
            "node_group_template_id": "${SPARK_MASTER_GROUP_ID}"
        },
        {
            "count": 1,
            "name": "${NAME}-worker",
            "node_group_template_id": "${SPARK_WORKER_GROUP_ID}"
        }
    ],
    "plugin_name": "spark",
    "shares": [],
    "use_autoconfig": true
}
EOF

echo $NAME-cluster-template:
cat spark-cluster-template.json
sahara cluster-template-create --json spark-cluster-template.json

创建集群:

# Create cluster
CLUSTER_TEMPLATE_ID=$(sahara cluster-template-list 2>/dev/null | grep -P "\s${NAME}-cluster-template\s" | awk '{print $4}')
cat >spark_cluster.json <<EOF
{
    "cluster_template_id": "${CLUSTER_TEMPLATE_ID}",
    "default_image_id": "${IMAGE_ID}",
    "description": "",
    "hadoop_version": "1.6.0",
    "is_protected": false,
    "is_public": true,
    "name": "${NAME}",
    "neutron_management_network": "${NETWORK_ID}",
    "plugin_name": "spark",
    "user_keypair_id": "${KEYPAIR_NAME}"
}
EOF

echo $NAME:
cat spark_cluster.json
sahara cluster-create --json spark_cluster.json

查看集群列表:

sahara cluster-list

完整脚本地址: create-spark-cluster.sh

提交spark任务

初始化变量:

# 任务名称
JOB_NAME=int32bit-spark-wordcount.jar
# jar包路径
BINARY_FILE=spark-wordcount.jar
# input hdfs路径
INPUT=hdfs://int32bit-spark-spark-master-0:8020/user/ubuntu/hello.txt
# output hdfs 路径
OUTPUT=hdfs://int32bit-spark-spark-master-0:8020/user/ubuntu/output2
# 集群的uuid
CLUSTER_UUID=27d3fc76-d913-4d64-8491-5fc0968efe90
# main函数所在的class
MAIN_CLASS=sahara.edp.spark.SparkWordCount

注意:

• INPUT的测试文件需要先上传到hdfs中。
• OUTPUT指定的hdfs路径不能存在

下载并上传jar包到sahara中:

# Upload job binary data
wget 'https://github.com/int32bit/openstack-cheat-sheet/raw/master/sahara/spark-wordcount.jar' -O ${BINARY_FILE}
sahara job-binary-data-create --name "${JOB_NAME}" --file "${BINARY_FILE}"

# Create job binary
DATA_ID=$(sahara job-binary-data-list 2>/dev/null | grep -P -- "\s${JOB_NAME}\s" | awk '{print $2}')
sahara job-binary-create --name "${JOB_NAME}" --url "internal-db://${DATA_ID}"

创建job模板:

# Create job template
BINARY_ID=$(sahara job-binary-list | grep -P -- "\s${JOB_NAME}\s" | awk '{print $2}')
cat >spark-job-template.json <<EOF
{
    "description": "",
    "interface": [],
    "is_protected": false,
    "is_public": true,
    "libs": [],
    "mains": [
        "${BINARY_ID}"
    ],
    "name": "${JOB_NAME}-template",
    "type": "Spark"
}
EOF
sahara job-template-create --json spark-job-template.json

提交job:

# Submit job
JOB_TEMPLATE_UUID=$(sahara job-template-list 2>/dev/null | grep -P -- "\s${JOB_NAME}-template\s" | awk '{print $2}')
cat >job.json <<EOF
{
    "cluster_id": "${CLUSTER_UUID}",
    "is_protected": false,
    "is_public": true,
    "interface": {},
    "job_configs": {
        "args": [
            "${INPUT}",
            "${OUTPUT}"
        ],
        "configs": {
            "edp.hbase_common_lib": true,
            "edp.java.java_opts": "",
            "edp.java.main_class": "${MAIN_CLASS}",
            "edp.spark.adapt_for_swift": true,
            "edp.substitute_data_source_for_name": true,
            "edp.substitute_data_source_for_uuid": true
        }
    }
}
EOF
sahara job-create --job-template "${JOB_TEMPLATE_UUID}" --json job.json

查看job状态:

sahara job-list

完整脚本: submit-spark-job.sh

其它链接

1. sahara测试: https://github.com/openstack/sahara-tests

manila

magnum