When you deploy a wsp solution in a sharepoint farm, the solution is first deployed on the server you are running the stsadm, then a timer job starts to deploy the solution files to each front end (in your case server 2 and 3)
It could be that the solution is not successfully deployed on server 2 and 3, due deployment failure
MSDN:
During deployment to the front-end Web servers, the following failures can occur:
If the timer service is not activated on a front-end Web server, then the deployment job remains stopped. On the pending jobs page in the user interface, the job appears as pending but not being serviced. The administrator either must fix the timer service, or cancel the deployment job.
If the Windows SharePoint Services Administrative service is not activated on a particular computer, an error code is set in the SPRunningJob object that marks the stage as failed and prevents any further operation. The failed deployment is converted to an administrative alert that notifies the administrator that the job failed due to a Windows SharePoint Services Administrative Web service that was not running.
If the extraction of a solution package (.wsp) fails on any particular server, the stage is marked as having failed and processing stops.
http://msdn.microsoft.com/en-us/library/aa544500.aspx
If one or more files fails to copy—for example, an existing file is marked as read-only—then the stage is marked as failed, and processing stops.