We have had a problem at ClearPointe getting an Operations Manager agent to install on one particular remotely managed server. Attempts to push an agent with the OM Discovery Wizard would fail with the following error:

One or more computers you are trying to manage are already in the process of being managed. Please resolve these issues via the Pending Management view in Administration, prior to attempting to manage them again.

The server does not appear in Pending Management of the Administration space, nor anywhere else in the Operations Manager Console for that matter. 

After making one final attempt to push an agent today, I decided to note the error message and place a post on the Microsoft Partners newsgroup.  Before logging onto the newsgroup site, I tried one last Bing search on the error message.  I came across a post on Kevin Holman’s OpsMgr Blog that I had not seen previously and it described our issue exactly.

Pending action information is stored in the agentpendingaction table of the Operations Manager database and it is possible for this table to get out of sync with the console.  Normally, this will happen if an agent install or upgrade process crashes or is terminated before it completes.  Subsequent attempts to install an agent fail with the above error due to this synchronization issue.

The first troubleshooting step is to check the agentpendingaction table of the Operations Manager database for the presence of a row that references the “ghost” pending action.  Run the following SQL query against the Operations Manager database:

select * from agentpendingaction

Next use the Operations Manager powershell to attempt to reject the “ghost” pending action.  Open the command shell with elevated privileges and run the following, replacing servername.domain.com with the appropriate FQDN:

get-agentPendingAction | where {$_.AgentName -eq “servername.domain.com”}|Reject-agentPendingAction

At this point, an exception might be returned in the command shell, which is exactly what happened in my case.  There is one last resort, albeit not supported, so run this at your own risk.  There is a stored procedure in SQL that will delete pending actions.  Run the following query against the Operations Manager database:

exec p_AgentPendingActionDeleteByAgentName ‘agentname.domain.com’

Again, substitute the appropriate agent name as it appears in the SQL table in place of agentname.domain.com.  The stored procedure worked for me without issue.  I was then able to push an agent to the remote server using the Discovery Wizard and am happy to report that the remote server currently has a status of “Healthy” in the Operations Manager Console.  Thank you again, Kevin Holman!