Build a cloud computing platform ~ Eucalyptus Installation Log centos 5.3

I just want to keep a journal account, not a technical document.

1. First find document http://open.eucalyptus.com/wiki/EucalyptusPrerequisites_v1.5.2

Check what needs to be prepared before installation.

2. Go http://open.eucalyptus.com/downloads

Download an installation package. I am using centos5.3

3. decompress the package and enter the decompressed directory. This step can be done later.

4. Add various packages that may be required:

// To synchronize with the clock <br/> Yum install-y NTP <br/> ntpdate pool.ntp.org <br/> // install the xen VM <br/> Yum install-y xen <br/> SED -- In-Place's/# (xend-http-server no) /(xend-http-server yes)/'/etc/xen/xend-config.sxp <br/> SED -- In-Place's/# (xend-address localhost) /(xend-address localhost)/'/etc/xen/xend-config.sxp <br/>/etc/init. d/xend restart <br/> // to make eucalyptus run properly <br/> Yum install-y java-1.6.0-openjdk ant-nodeps DHCP bridge-utils <br/>

Then open the firewall port. Input: System-config-securitylevel. I only open ports 8443,877 3 and 8774,877 5. It is said that the first three ports of the front-end must be enabled. 8775 is required for node. However, eucalyptus recommends that you shut down all the firewalls. It can be seen that this project is only implemented

But the security of the cloud is not considered yet.

5. The command to install the front-end is as follows: (the front-end refers to a PC dealing with users, and the node can be regarded as a node in the cluster). x86_64 in the following command can be replaced. I used 586 during installation. In the cloud installation command, some x86_64 changed to i8.0. However, no matter what. It seems that you can install all the RPM packages in it.

// Install third-party software <br/> Cd eucalyptus-1.5.2-rpm-deps-x86_64 <br/> rpm-uvh aoetools-21-1.el4.x86_64.rpm/<br/> euca-axis2c-1.5.0-2.x86_64.rpm/<br/> euca-rampartc-1.2.0-1.x86_64.rpm/<br/> vblade-14-1mdv2008.1.x86_64.rpm <br/>/> CD .. <br/> // install cloud <br/> rpm-uvh eucalyptus-1.5.2-1.x86_64.rpm/<br/> eucalyptus-cloud-1.5.2-1.x86_64.rpm/<br/> eucalyptus-gl-1.5.2-1.x86_64.rpm/<br/> eucalyptus-cc-1.5.2-1.x86_64.rpm <br/>

6. Package to be installed by Node

CD eucalyptus-1.5.2-rpm-deps-x86_64 <br/> rpm-uvh aoetools-21-1.el4.x86_64.rpm/<br/> euca-axis2c-1.5.0-2.x86_64.rpm/<br/> euca-rampartc-1.2.0-1.x86_64.rpm/<br/> euca-libvirt-1.5-1.x86_64.rpm/<br/> vblade-14-1mdv2008.1.x86_64.rpm <br/>

// Install Node Controller <br/> rpm-uvh eucalyptus-1.5.2-1.x86_64.rpm/<br/> eucalyptus-gl-1.5.2-1.x86_64.rpm/<br/> eucalyptus-nc-1.5.2-1.x86_64.rpm <br/>

7. I am here for convenience. Place node and front-end on a PC. I don't know if it is feasible. Just try again. After completing step 5, return

Eucalyptus-1.5.2-rpm-deps-i586, install the euca-libvirt-1.5-1.i386.rpm, and then go back to the parent directory to install the eucalyptus-nc-1.5.2-1.i386.rpm

8. Then enter su eucalyptus-c "virsh list" according to the instructions. It is true that fail to connect xend, but it does not show domain-0 as it is in the legend.

9. Temporarily ignore it. The frontend must execute

/Etc/init. d/eucalyptus-cloud start <br/>/etc/init. d/eucalyptus-CC start <br/>

10. Run/etc/init. d/eucalyptus-NC start on the node.

11. For me to perform these three operations on the same PC. When executing cloud, you should say that you should have at least 32 loop devices. You can run the following command: ls-L/dev/loop * | WC-L to check whether you have several loop devices. if it is less than 32, VIM/etc/modprobe. conf (different machines may be different here), add a line: Options loop max_loop = 64, and then modprobe loop. Input LS-L/dev/loop * | WC-L to check whether the value is 64.

If it is 64, continue typing the command. An error may be reported during NC startup:

You shoshould have at least 32 loop Devices
Starting eucalyptus services: Node Controller cannot be started: errors in/opt/eucalyptus/var/log/eucalyptus/euca_test_nc.log

I have checked the following/opt/eucalyptus/var/log/eucalyptus/euca_test_nc.log file. There are the following lines.

 

Libvir: xen daemon error: Internal error failed to connect to xend
Libvir: xen daemon error: Unable to connect to 'localhost: 100': Connection refused
Libvirt error: Unable to connect to 'localhost: 8080': Connection refused (code = 38)

Is port 8000 not enabled? Close the firewall and SELinux, and try again. The result is still this error. What software is not installed? Try this command again:

Yum install-y java-1.6.0-openjdk-devel ant-nodeps libvirt-devel curl-devel/
Httpd-devel Apr-devel OpenSSL-devel DHCP
The result still does not work. It does not appear that the package is not installed. Run the su eucalyptus-c "virsh list" command to obtain the "libvir: Error: operation failed:
Xenproxyopen error: failed to connect to the hypervisor. The xend service may not be viewed. An exception is thrown when xend start is used.
(You cannot use system-config-services to check whether xend has been started. I checked the result and checked it for a long time .) How to enable it?
First, restart. Grub will display a new line (SELinux must be disabled before restart, and the method is the second tab of system-config-securitylevel,
Select disable. Otherwise, it takes a long time to enter the xen .). Enter xen. Xend start first, and XM list can be used. This time you should be able to see
That domain-0. Then you can start it with the eucalyptus command. However, you may also receive a warning for loop device less than 32.
Change the/etc/modprobe. conf file again, add options loop max_loop = 64, and then enter modprobe loop in the command line.
Run the following command:
Eucalyptus-cloud start
Eucalyptus-CC start
Eucalyptus-NC start
Is Cloud alread start? . Use eucalyptus-cloud restart if you are not at ease. There should be no errors this time!
Installation is complete! (I only used one machine to install cloud... I haven't tried it yet .)

If the following error occurs:

Restarting eucalyptus cluster controller: (13) Permission denied: make_sock: cocould not bind to address [::]: 8774
(13) Permission denied: make_sock: cocould not bind to address 0.0.0.0: 8774
No listening sockets available, shutting down
Unable to open logs
Failed to start the CC!

It must be because the firewall is not set to disable or SELinux is not set to passivate or disable in system-config-security.