07.计算Nova→3.场景学习→2.Launch→3.nova-compute

返回总目录

描述

详细

为 instance 准备资源

nova-compute 首先会根据指定的 flavor 依次为 instance 分配内存、磁盘空间和 vCPU。
网络资源也会提前分配。

Jun 16 13:12:52 compute nova-compute[5666]: 
DEBUG nova.compute.manager 
[None req-d9a23179-e16d-4016-9586-e68c3dfdf6a9 admin admin] 
[instance: 86f71e87-4ae2-48fb-9bcd-c72e3c28e093] Starting instance... 
{{(pid=5666) _do_build_and_run_instance /opt/stack/nova/nova/compute/manager.py:1835

Jun 16 13:12:52 compute nova-compute[5666]: 
DEBUG oslo_concurrency.lockutils 
[None req-d9a23179-e16d-4016-9586-e68c3dfdf6a9 admin admin] 
Lock "compute_resources" acquired 
by "nova.compute.resource_tracker.instance_claim" :: waited 0.000s 
{{(pid=5666) inner 
/usr/local/lib/python2.7/dist-packages/oslo_concurrency/lockutils.py:273

Jun 16 13:12:52 compute nova-compute[5666]: 
DEBUG nova.compute.resource_tracker 
[None req-d9a23179-e16d-4016-9586-e68c3dfdf6a9 admin admin] 
Memory overhead for 512 MB instance; 0 MB 
{{(pid=5666) instance_claim /opt/stack/nova/nova/compute/resource_tracker.py:202

(ADJ) 在头顶上的；在上面的；高架的→You use overhead to indicate that something is above you or above the place that you are talking about.

n. 经常费用; 经常开支;

这里指开销

Jun 16 13:12:52 compute nova-compute[5666]: 
DEBUG nova.compute.resource_tracker 
[None req-d9a23179-e16d-4016-9586-e68c3dfdf6a9 admin admin] 
Disk overhead for 1 GB instance; 0 GB 
{{(pid=5666) instance_claim /opt/stack/nova/nova/compute/resource_tracker.py:205

Jun 16 13:12:52 compute nova-compute[5666]: 
DEBUG nova.compute.resource_tracker 
[None req-d9a23179-e16d-4016-9586-e68c3dfdf6a9 admin admin] 
CPU overhead for 1 vCPUs instance; 0 vCPU(s) 
{{(pid=5666) instance_claim /opt/stack/nova/nova/compute/resource_tracker.py:208

Jun 16 13:12:52 compute nova-compute[5666]: 
INFO nova.compute.claims 
[None req-d9a23179-e16d-4016-9586-e68c3dfdf6a9 admin admin] 
[instance: 86f71e87-4ae2-48fb-9bcd-c72e3c28e093] 
Attempting claim on node compute: memory 512 MB, disk 1 GB, vcpus 1 CPU

创建虚拟机时选择的flavor是

Jun 16 13:12:52 compute nova-compute[5666]: 
INFO nova.compute.claims 
[None req-d9a23179-e16d-4016-9586-e68c3dfdf6a9 admin admin] 
[instance: 86f71e87-4ae2-48fb-9bcd-c72e3c28e093] 
Total memory: 7983 MB, used: 512.00 MB

Jun 16 13:12:52 compute nova-compute[5666]: 
INFO nova.compute.claims 
[None req-d9a23179-e16d-4016-9586-e68c3dfdf6a9 admin admin] 
[instance: 86f71e87-4ae2-48fb-9bcd-c72e3c28e093] 
memory limit not specified, defaulting to unlimited

Jun 16 13:12:52 compute nova-compute[5666]: 
INFO nova.compute.claims 
[None req-d9a23179-e16d-4016-9586-e68c3dfdf6a9 admin admin] 
[instance: 86f71e87-4ae2-48fb-9bcd-c72e3c28e093] Total disk: 47 GB, used: 0.00 GB

Jun 16 13:12:52 compute nova-compute[5666]: 
INFO nova.compute.claims 
[None req-d9a23179-e16d-4016-9586-e68c3dfdf6a9 admin admin] 
[instance: 86f71e87-4ae2-48fb-9bcd-c72e3c28e093] 
disk limit not specified, defaulting to unlimited

Jun 16 13:12:52 compute nova-compute[5666]: 
INFO nova.compute.claims 
[None req-d9a23179-e16d-4016-9586-e68c3dfdf6a9 admin admin] 
[instance: 86f71e87-4ae2-48fb-9bcd-c72e3c28e093] Total vcpu: 4 VCPU, used: 0.00 VCPU

4VCPU是因为该日志是在公司开发环境搭建的2节点devstack上产生的，每台机器VCPU数量为4

Jun 16 13:12:52 compute nova-compute[5666]: 
INFO nova.compute.claims 
[None req-d9a23179-e16d-4016-9586-e68c3dfdf6a9 admin admin] 
[instance: 86f71e87-4ae2-48fb-9bcd-c72e3c28e093] 
vcpu limit not specified, defaulting to unlimited

Jun 16 13:12:52 compute nova-compute[5666]: 
DEBUG nova.virt.hardware 
[None req-d9a23179-e16d-4016-9586-e68c3dfdf6a9 admin admin] 
Require both a host and instance NUMA topology to fit instance on host. 
{{(pid=5666) numa_fit_instance_to_host /opt/stack/nova/nova/virt/hardware.py:1465

Jun 16 13:12:52 compute nova-compute[5666]: 
INFO nova.compute.claims 
[None req-d9a23179-e16d-4016-9586-e68c3dfdf6a9 admin admin] 
[instance: 86f71e87-4ae2-48fb-9bcd-c72e3c28e093] Claim successful on node compute

网络资源也会提前分配。

Jun 16 13:12:53 compute nova-compute[5666]: 
DEBUG nova.compute.manager 
[None req-d9a23179-e16d-4016-9586-e68c3dfdf6a9 admin admin]
 [instance: 86f71e87-4ae2-48fb-9bcd-c72e3c28e093] 
Allocating IP information in the background. 
{{(pid=5666) _allocate_network_async /opt/stack/nova/nova/compute/manager.py:1423

创建 instance 的镜像文件

首先将该 image 下载到计算节点，然后将其作为 backing file 创建 instance 的镜像文件。

从 Glance 下载 image

资源准备好之后，nova-compute 会为 instance 创建镜像文件。OpenStack 启动一个 instance 时，会选择一个 image，这个 image 由 Glance 管理。

nova-compute 首先会检查 image 是否已经下载（比如之前已经创建过基于相同 image 的 instance）。如果没有，就从 Glance 下载 image 到本地。由此可知，如果计算节点上要运行多个相同 image 的 instance，只会在启动第一个 instance 的时候从 Glance 下载 image，后面的 instance 启动速度就大大加快了。

日志分析：
image（ID为 f0bdf0e7-ebb5-459d-a7a5-a0f0ed13d6d7）是 qcow2 格式，nova-compute 将其下载。Nova 默认会通过 qemu-img 转换成 raw 格式，以提高 IO 性能。
下载的image 的存放目录是 /opt/stack/data/nova/instances/_base，这是由/etc/nova/nova.conf 的两个配置选项决定的
1
2
3
4
5
[DEFAULT]
...
instances_path = /opt/stack/data/nova/instances
base_dir_name = _base
...
下载的 image 文件被命名为 6eb03a619bcf59de5a27bf520cf83606bf8327a7，这是 image id 的 SHA1 哈希值Secure Hash Algorithm

Jun 16 13:12:55 compute nova-compute[5666]: 
DEBUG nova.virt.images 
[None req-d9a23179-e16d-4016-9586-e68c3dfdf6a9 admin admin] 
f0bdf0e7-ebb5-459d-a7a5-a0f0ed13d6d7 was qcow2, converting to raw 
{{(pid=5666) fetch_to_raw /opt/stack/nova/nova/virt/images.py:179

Jun 16 13:12:55 compute nova-compute[5666]: 
DEBUG nova.utils 
[None req-d9a23179-e16d-4016-9586-e68c3dfdf6a9 admin admin] 
Path '/opt/stack/data/nova/instances' supports direct I/O 
{{(pid=5666) supports_direct_io /opt/stack/nova/nova/utils.py:1394

Jun 16 13:12:55 compute nova-compute[5666]: 
DEBUG oslo_concurrency.processutils 
[None req-d9a23179-e16d-4016-9586-e68c3dfdf6a9 admin admin] 
Running cmd (subprocess): qemu-img convert -t none -O raw -f qcow2 
/opt/stack/data/nova/instances/_base/6eb03a619bcf59de5a27bf520cf83606bf8327a7.part 
/opt/stack/data/nova/instances/_base/6eb03a619bcf59de5a27bf520cf83606bf8327a7.converted 
{{(pid=5666) execute /usr/local/lib/python2.7/dist-packages/oslo_concurrency/processutils.py:372

-O：output format；-f：format；-t：cache

backing file的id不一样是因为这条是我后来添加的笔记(又一个实例)

为 instance 创建镜像文件
有了下载的image 之后，instance 的镜像文件直接通过 qemu-img 命令创建，backing file 就是下载的 image。
这里 instance 的镜像文件是 /opt/stack/data/nova/instances/86f71e87-4ae2-48fb-9bcd-c72e3c28e093/disk，格式为 qcow2，其中 86f71e87-4ae2-48fb-9bcd-c72e3c28e093就是 instance的 id。可以通过 qume-info 查看 disk 文件的属性

Jun 16 13:12:57 compute nova-compute[5666]: 
DEBUG oslo_concurrency.processutils 
[None req-d9a23179-e16d-4016-9586-e68c3dfdf6a9 admin admin] 
Running cmd (subprocess): qemu-img create -f qcow2 -o 
backing_file=/opt/stack/data/nova/instances/_base/6eb03a619bcf59de5a27bf520cf83606bf8327a7 
/opt/stack/data/nova/instances/86f71e87-4ae2-48fb-9bcd-c72e3c28e093/disk 
{{(pid=5666) execute /usr/local/lib/python2.7/dist-packages/oslo_concurrency/processutils.py:372

-o：options

root@compute:~# qemu-img info /opt/stack/data/nova/instances/86f71e87-4ae2-48fb-9bcd-c72e3c28e093/disk
image: /opt/stack/data/nova/instances/86f71e87-4ae2-48fb-9bcd-c72e3c28e093/disk
file format: qcow2
virtual size: 1.0G (1073741824 bytes)
disk size: 1.9M
cluster_size: 65536
backing file: /opt/stack/data/nova/instances/_base/6eb03a619bcf59de5a27bf520cf83606bf8327a7
Format specific information:
    compat: 1.1
    lazy refcounts: false
    refcount bits: 16
    corrupt: false

说明：
image，指的是 Glance 上保存的镜像，作为 instance 运行的模板。计算节点将下载的 image 存放在 /opt/stack/data/nova/instances/_base 目录下。
镜像文件，指的是 instance 启动盘所对应的文件
二者的关系是：image 是镜像文件的 backing file。image 不会变，而镜像文件会发生变化。比如安装新的软件后，镜像文件会变大。
英文中两者都叫 “image”，为避免混淆，我们用 “image” 和 “镜像文件” 作区分。

创建 instance 的 XML 定义文件

cloudman：创建的 XML 文件会保存到该 instance 目录 /opt/stack/data/nova/instances/f1e22596-6844-4d7a-84a3-e41e6d7618ef，命名为 libvirt.xml
我：在/opt/stack/data/nova/instances/86f71e87-4ae2-48fb-9bcd-c72e3c28e093/没有找到.xml这个文件

但是在计算节点的/etc/libvirt/qemu/发现xml文件，该文件记录了虚机的各种参数

cloudman：

我的：
日志中关于xml的有：

1
2
3

DEBUG nova.virt.libvirt.driver
[instance: 86f71e87-4ae2-48fb-9bcd-c72e3c28e093] 
Start _get_guest_xml network_info=[...

1
2
3

DEBUG nova.virt.libvirt.driver
[instance: 86f71e87-4ae2-48fb-9bcd-c72e3c28e093] 
End _get_guest_xml ...

这个过程应该是获取xml文件吧，然后正式启动创建过程

root@compute:~# ll /opt/stack/data/nova/instances/86f71e87-4ae2-48fb-9bcd-c72e3c28e093/
total 2052
drwxr-xr-x 2 stack libvirt    4096 Jun 16 13:12 ./
drwxr-xr-x 5 stack root       4096 Jun 16 13:12 ../
-rw------- 1 root  root      48807 Jun 16 15:23 console.log
-rw-r--r-- 1 root  root    2097152 Jun 16 15:23 disk
-rw-r--r-- 1 stack libvirt      86 Jun 16 13:12 disk.info

创建虚拟网络并启动 instance

创建虚拟网络：用的是 linux-bridge 实现的虚拟网络

cloudman：为 instance 创建虚拟网络设备
我：没有找到Ensuring bridge

cloudman：

vif是virtual interface(虚拟网卡)的缩写
我：

Jun 16 13:12:57 compute nova-compute[5666]: 
DEBUG nova.network.base_api 
[None req-d9a23179-e16d-4016-9586-e68c3dfdf6a9 admin admin] 
[instance: 86f71e87-4ae2-48fb-9bcd-c72e3c28e093] 
Updating instance_info_cache with network_info: 
[{"profile": {}, ... "network": {"bridge": "brq4f28b3ec-cc", 
"subnets": [{"ips": [{"meta": {}, "version": 4, ... "address": "172.12.12.7"}], 
"version": 4, "meta": {"dhcp_server": "172.12.12.2"}, ... "cidr": "172.12.12.0/24", ...
 {{(pid=5666) update_instance_cache_with_nw_info /opt/stack/nova/nova/network/base_api.py:48

虽然没有Ensuring bridge，但是有nova.virt.libvirt.vif参考见07.计算Nova→3.通过场景学习nova→4.Start Instance。总之，上述过程表示网络相关

接下来可以启动 instance了，最后Instance spawned successfully表示实例成功孵化。
OpenStack 图形界面和 KVM CLI 都可以查看到 instance 的运行状态。
1.virsh list用于显示当前节点创建的实例，而不是像nova list似的显示所有节点创建的实例 2.由于是第一次创建实例，故Name是...1

1
2
3

Jun 16 13:13:01 compute nova-compute[5666]: 
INFO nova.compute.manager [-] 
[instance: 86f71e87-4ae2-48fb-9bcd-c72e3c28e093] VM Started (Lifecycle Event)

Jun 16 13:13:01 compute nova-compute[5666]: 
INFO nova.compute.manager 
[None req-9b9e116b-fa49-46b3-8c41-9bade624c334 None None] 
[instance: 86f71e87-4ae2-48fb-9bcd-c72e3c28e093] VM Paused (Lifecycle Event)

Jun 16 13:13:01 compute nova-compute[5666]: 
INFO nova.compute.manager 
[None req-9b9e116b-fa49-46b3-8c41-9bade624c334 None None] 
[instance: 86f71e87-4ae2-48fb-9bcd-c72e3c28e093] 
During sync_power_state the instance has a pending task (spawning). Skip.

同步电源状态

Jun 16 13:13:01 compute nova-compute[5666]: 
INFO nova.compute.manager 
[None req-9b9e116b-fa49-46b3-8c41-9bade624c334 None None] 
[instance: 86f71e87-4ae2-48fb-9bcd-c72e3c28e093] VM Resumed (Lifecycle Event)

resume [rɪ'zjum](V-ERG) （使）重新开始;（使）继续进行

1
2
3

Jun 16 13:13:01 compute nova-compute[5666]: 
INFO nova.virt.libvirt.driver [-] 
[instance: 86f71e87-4ae2-48fb-9bcd-c72e3c28e093] Instance spawned successfully.

spawn [spɔn]：孵化

posted @ 2019-07-24 16:13 LeisureZhao 阅读(183) 评论(0) 编辑收藏举报

刷新页面返回顶部