If AM waits for hours for an updated hashrate from ccminer that it has connection issues or it is not responding, and does not take time into consideration, it is not the correct behavior.
AM should have a timer that checks every x seconds the reported hashrate from the miner's API, with one minute delay when miner is started to allow all GPUs to be started and send hash rates. Like this if one pool / algo server is offline, it will change to the next most profitable algo from the list. RIGHT ?
If this happens often to you, you should setup or adjust a rule based upon Accepted progress. If the accepted rate doesn't increase after X number of minutes, then have AM apply a template (or you could do another pool) so it will mine something else.