Since the only error that actually reaches here is a user error (not putting an inference server container in the launcher Pod template), the proper handling of this error is to, in addition to ignoring the LauncherConfig here, expose the error to users in the Status section of the LauncherConfig (which, I see, does not yet have any place to report problems). Can be handled in other PR(s).
Originally posted by @MikeSpreitzer in #374 (comment)
Since the only error that actually reaches here is a user error (not putting an inference server container in the launcher Pod template), the proper handling of this error is to, in addition to ignoring the LauncherConfig here, expose the error to users in the
Statussection of theLauncherConfig(which, I see, does not yet have any place to report problems). Can be handled in other PR(s).Originally posted by @MikeSpreitzer in #374 (comment)