Cluster of vCD Cells

[USPanda]

Hi,

How to configure cluster between 2 vCD 5.1.2 Cells installed on RHEL 6.3 x64

Is there any document available for the same (can someone please share the link or steps to configure it)

Thanks,

[IamTHEvilONE]

The installation guide has two parts.  Installing a single cell, then adding additional cells.

If you run more than 1 cell, you will require an NFS mount for the cells to share as temporary storage.

In addition, you should have some front end load balancer for the web and console proxy services.

See for features: VMware KB 1026294

Installation: http://pubs.vmware.com/vcd-51/topic/com.vmware.ICbase/PDF/vcd_51_install.pdf - Page 23 "Creating a vCloud Director Server Group"

[USPanda]

Thanks for the details, which is really helpful. However I have tried this guide for section "Install vCloud Director Software on Additional Servers" but it didn't work out for me. Possibly I have already executed the same command without -r in Cell2. So any alternate work around at this stage?

1. copied the "responses.properties" file from Cell1 to Cell2 & the path is "/opt/vmware/vcloud-director/bin/"

2. in Cell2 --> # ls -ltrh /opt/vmware/vcloud-director/bin/responses.properties

                          -rw-------. 1 vcloud vcloud   430 Jul  9 12:26 responses.properties

                       # cd /opt/vmware/vcloud-director/bin/

2. in Cell2 I have tried to configure with -r option

[root@vcd5bin2 tmp]# ./vmware-vcloud-director-5.1.2-1068441.bin -r /opt/vmware/vcloud-director/etc/responses.properties
Before you upgrade to vCloud Director 5.1.2, read VMware KB article 2047922, which provides tools and procedures that enable you to detect and correct network adapter type mismatches before you upgrade. If you upgrade before correcting these mismatches, some of your virtual machines might have no network connectivity after the upgrade. Do you wish to proceed with the upgrade? (y/n)?y
Continuing installation.
Checking architecture...done
Checking for a supported Linux distribution...Detected Red Hat Linux system
done
Checking for necessary RPM prerequisites...done
Checking free disk space...done
Extracting VMware vCloud Director. Please wait, this could take a few minutes...
vmware-vcloud-director-5.1.2-1068441.x86_64.rpm
vmware-vcloud-director-rhel-5.1.2-1068441.x86_64.rpm
done

WARNING: the RPMs for vCloud Director are already installed, skipping
installation.

If you have not already run the configuration script or
would like to reconfigure the product the command to run is:
/opt/vmware/vcloud-director/bin/configure

[USPanda]

already tried to remove as well, but still some error... not sure if its related to DB timeout or something else !!

[root@vcd5bin2 ~]# rpm -e vmware-vcloud-director

Stopping vmware-vcd-watchdog: [  OK  ]

Stopping vmware-vcd-cell: Unable to stop cell[FAILED]

ERROR: failed to stop services

error: %preun(vmware-vcloud-director-5.1.2-1068441.x86_64) scriptlet failed, exit status 1

[root@vcd5bin2 ~]# tail /opt/vmware/vcloud-director/logs/cell.log

Application startup begins: 7/18/13 5:43 PM

Successfully bound network port: 80 on host address: 172.22.6.140

Successfully bound network port: 443 on host address: 172.22.6.140

Successfully connected to database: jdbc:jtds:sqlserver://172.22.6.35:1433/vcloud;socketTimeout=90

Current locale "en" verified successfully.

Successfully bound network port: 61616 on host address: 172.22.6.140

Successfully bound network port: 61613 on host address: 172.22.6.140

Uncaught Exception. Originating thread: Thread[SpringOsgiExtenderThread-4,5,spring-osgi-extender[67673a2]-threads]. Message: Error creating bean with name 'globalSessionFactoryService': Invocation of init method failed; nested exception is org.springframework.beans.factory.BeanCreationException: Error creating bean with name 'globalSessionFactory' defined in URL [bundleentry://189.fwk19627754/META-INF/spring/common-persist-context.xml]: Cannot create inner bean 'net.sf.ehcache.hibernate.SingletonEhCacheProvider#23ec4333' of type [net.sf.ehcache.hibernate.SingletonEhCacheProvider] while setting bean property 'cacheProvider'; nested exception is org.springframework.beans.factory.BeanCreationException: Error creating bean with name 'hibernateCacheRegions': Cannot resolve reference to bean 'configModelCache' while setting bean property 'sourceList' with key [0]; nested exception is org.springframework.beans.factory.BeanCreationException: Error creating bean with name 'configModelCache' defined in URL [bundleentry://189.fwk19627754/META-INF/spring/common-persist-context-cache.xml]: Cannot resolve reference to bean 'cacheManager' while setting bean property 'cacheManager'; nested exception is org.springframework.beans.factory.BeanCreationException: Error creating bean with name 'cacheManager' defined in URL [bundleentry://189.fwk19627754/META-INF/spring/common-persist-context.xml]: Invocation of init method failed; nested exception is net.sf.ehcache.CacheException: Problem creating connections: Error decrypting data.

[admin]

The crucial piece of information in the responses file that needs to be conveyed from the initial cell to additional members of the server group is the private key that the cluster will use to encrypt/decrypt data in both properties files as well as the database. If all members of the server group aren't using the same key, you'll encounter decryption errors.

The most straightforward way to get the additional cells working would be to remove the RPM if it's not already gone and to remove anything under /opt/vmware/vcloud-director/etc to ensure that it doesn't use the old properties files from the bad install. Then re-run the installer and supply the -r option and specify the path to the responses.properties file you've copied over. If the RPM is already installed but you haven't run the configure tool, you can start from that point ($VCLOUD_HOME/bin/configure -r /path/to/responses.properties -- that's all the installer does when it prompts you).

Note that for the permissions on responses.properties; if you've already installed the product the vcloud user and group will already exist (getent /etc/passwd vcloud to check) so the permissions you mentioned should be adequate. If the vcloud user does not (or no longer) exists you'll need to set it to be more permissive -- most likely the read bit will need to be set for other temporarily. i.e. chmod o+r responses.properties before running the installer/configure tool and then remove it when done -- chmod o-r responses.properties.

[USPanda]

Thanks for the details. I have tried to uninstall the RPM but its not happening. Possibly need to manually remove the folder as you suggested.

Also what is the role of NFS share (where to use) if we are going to use 2 cells in a group. I didn't notice this in installation guide. But saw in forums.

[admin]

You'll need to remove the RPM before attempting to run the installer. You can terminate any of the vCD processes using kill -9 <pid>. Just identify any running vCD processes (e.g. ps auxw | grep vcd) and then kill them. Once the processes aren't running it should be possible to erase the RPM. If it continues to fail to uninstall once you've killed the processes you can add the --noscripts flag to have rpm skip the failing part of the uninstall process (see the link above for more details about what it means).

[USPanda]

this was a fresh installation for Cell2 to join with Cell1 to make a group for high availability. So Cell2 is not is use so far.

[root@vcd5bin2 ~]# ps aux |grep vcd

root      3925  0.0  0.0 103300   848 pts/0    S+   19:59   0:00 grep vcd

[root@vcd5bin2 ~]# rpm -e vmware-vcloud-director

Stopping vmware-vcd-watchdog: [  OK  ]

Stopping vmware-vcd-cell: Unable to stop cell[FAILED]

ERROR: failed to stop services

error: %preun(vmware-vcloud-director-5.1.2-1068441.x86_64) scriptlet failed, exit status 1

[root@vcd5bin2 ~]# /etc/init.d/vmware-vcd stop

Stopping vmware-vcd-watchdog:                              [  OK  ]

Stopping vmware-vcd-cell: Unable to stop cell              [FAILED]