This discussion has been locked. The information referenced herein may be inaccurate due to age, software updates, or external references.
You can no longer post new replies to this discussion. If you have a similar question you can start a new discussion in this forum.

HA Installation error

Hi,

We are now finally getting round to installing HA in our current Active-Active environment and have a couple of questions.

When running the HA installer on the server I wish to become the standby server I get the following errors and cannot get past these - despite the fact that this machine does have internet access.

I have tried downloading the installer from the customer portal & the main orion server and get the same issue each time.

HA-Error.JPG

Questions are -

1. Is there a correct way to convert a pair of existing active servers into a standby servers. (We have a Primary and Additional Polling engine)

2. Do I have to uninstall ALL of the software on the servers that I want to become the standby servers before running the HA Installer on them.

3. Do I have to build new servers completely and use these as my new Standby Servers.

Any help would be appreciated. aLTeReGo

  • dunky2k  wrote:

    1. Is there a correct way to convert a pair of existing active servers into a standby servers. (We have a Primary and Additional Polling engine)

    2. Do I have to uninstall ALL of the software on the servers that I want to become the standby servers before running the HA Installer on them.

    3. Do I have to build new servers completely and use these as my new Standby Servers.

    There is no way to 'convert' a server into an HA backup. Instead, it's best to start with a clean machine or revert the virtual machine back to its original state prior to the installation of another Orion product. Alternatively, you can uninstall all Orion products from the machine using the latest installer and running the following from the command prompt.

    'path to exe'\SolarWinds.Orion.Installer.exe -uninstall
  • Hi aLTeReGo

    Many thanks for the reply above.

    Now that we have restarted with clean machines i proceeded to follow the steps below

    1. Installed the HA installer on my Primary standby server that will protect my Primary poller and let the configuration wizard complete.

    2. Installed the HA installer on my Additional Standby Server that will protect my Primary Additional Poller and let the configuration wizard complete.

    Now when i go to the High Availability Deployment Summary page of my Main Primary Poller the status of these new servers are showing as critical (see below)

    181.JPG

    189.JPG

    When i go to the License Manager page of my main poller - i can see the HA Licenses albeit not assigned yet, suspect this is because i haven't created the pools yet, however i also see an entry for the additional polling engine for 189 which is supposed to be my standby additional polling server, should i see this, and if i am supposed to see this, then why cant i also see the standby for my primary poller? Have i perhaps taken the wrong option when installing the HA installer on the standby additional poller server?

    Lic161.JPG

    One final question, when i want to create the pools and it asks for the virtual hostname, does each pool have to have a unique name or since we have a Primary/Additional Poller acting as one instance does each pool just use the same hostname?

    I thought that i would have been able to group the Primary/Additional pollers together and fail these over to the standby group as a pair?

  • When you create the pool, does the status of the standby member change? The fact that you have a standby without an active member in a pool is enough to cause the status to be critical.

    dunky2k  wrote:

    When i go to the License Manager page of my main poller - i can see the HA Licenses albeit not assigned yet, suspect this is because i haven't created the pools yet, however i also see an entry for the additional polling engine for 189 which is supposed to be my standby additional polling server, should i see this, and if i am supposed to see this, then why cant i also see the standby for my primary poller? Have i perhaps taken the wrong option when installing the HA installer on the standby additional poller server?

    Lic161.JPG

    The license manager should sort itself out once the HA pools are created.

    One final question, when i want to create the pools and it asks for the virtual hostname, does each pool have to have a unique name or since we have a Primary/Additional Poller acting as one instance does each pool just use the same hostname?

    I thought that i would have been able to group the Primary/Additional pollers together and fail these over to the standby group as a pair?

    Unless each member of the pool are in different subnets, there is no requirement to provide a virtual hostname. This is completely optional in a same-subnet deployment. A VIP or Virtual Hostname is used to ensure inbound traffic is directed to the appropriate 'active' member.

    Currently, only two members are allowed to be in the same pool. An active and passive member.

  • aLTeReGo

    Many thanks for your continued support. the above information allowed me to look a little deeper into the issue and i discovered that my issue was some old 'orphaned' HA servers listed in the database. This came about as a result of how we did our failover processes in our then active-active environment.

    To get round it i cleaned up all of the orphaned records and uninstalled the HA software using the installer on both the Standby Main and Standby Additional pollers.

    A quick restart of the servers and a fresh install of the HA Installer again has now resolved that issue and i have now got HA running for now at the application level.

    Now onto the more trickier part, the database.

    Currently each SQL server has a standalone instance installed which serviced the active - active setup.

    What i intend to do now is as below -

    1. Create a SQL Failover Cluster installation on each of these SQL Severs.

    2. Shutdown the Orion services and dettach/backup the database from this instance.

    3. Re-attach/restore this database into the SQL Cluster Instance.

    4. Setup AlwaysOn Availability Groups / Listeners on the Database so that it can replicate to the other cluster instance.

    5. Restart the Orion services and run the configuration wizard to point to the SQL Listener for the Cluster instance rather than the standalone instance.

    6. Confirm that HA Application failover still functions as well as now SQL DB Failover.

    7. If all is working as expected uninstall the standaolne SQL instances from the Servers.

    I know that in an ideal world i would just create new servers, but for the time being this is not an option, so would the above steps in theory work, or is this a bad idea just waiting to go wrong.

  • The plan looks solid. If you're not already running SQL 2016 SP1 I would recommend also including a SQL upgrade in that process.

  • aLTeReGo​ yes we are already running SQL Server 2016 SP1

  • We experiencing similar issues after applying latest hotfix. The HA is broken and so far support has not been able to fix it, after around 4 weeks.

    The licensing keeps as 'not assigned'

    When trying to setup 'High Availability Pool' regardless the options the message is

    "Cannot create the HA pool. A pool must have both active and standby members"

    have tried editing OrionServers, HA_PoolMembers and engines tables and same outcome or no outcome (if deleting second entry in HA_PoolMembers). tried to validated against a old DB backup instance

    No visibility as of now