When the configuration store goes down, the IndexChangeManagerImpl functionality attempts to reinstate the connection almost every millisecond, and logs about the failure with stacktrace included. This can consume large amounts of disk space and makes the logs difficult to use.
Use the reproduction steps in OPENAM-14531, pyforge will shut down the configuration store (to build indexes) when AM is running, and the outcome is that the Policy debug log will grow to an unexpected size.
AM recognizes that the server is down, and applies delays when reconnecting to the server, similar backoff approach has been implemented elsewhere in AM already.
AM attempts to reestablish the connection every millisecond.