hmmm, have zero trouble to get CEP2 first time, all the time, but must say that I'm always using the restrict on -in the device profiles-, based on cache depth, have set 4 on laptop [default profile] and 7 on desktop [home profile], and in app_config set CEP2 to 1. The second project is OET1 on both. Because CEP2 is restricted to only allow 1 to run, the backlog is always oldest, so one CEP2 finishes there's an immediate take over by a new. All other cores [14] are occupied by OET. They have a limit of 7 in app_config, so in any case -even if CEP2 were not the oldest-, that one core is always doing clean energy.

Here's me sample app_config.xml from the wcg project folder:

<app_config>
   <app>
      <name>cep2</name>
      <max_concurrent>1</max_concurrent>
   </app>
   <app>
      <name>mcm1</name>
      <max_concurrent>6</max_concurrent>
      <fraction_done_exact/>
   </app>
   <app>
      <name>ugm1</name>
      <max_concurrent>2</max_concurrent>
      <fraction_done_exact/>
   </app>
   <app>
      <name>oet1</name>
      <max_concurrent>7</max_concurrent>
      <fraction_done_exact/>
   </app>
</app_config>

If neither comes, then the profiles are flipped to add MCM/UGM, where the devices are then usually swarmed by MCM and a few UGM, so set those to 6 and 2. Never an idle core here.

@ Techs (if there's a point at all directing suggestions at techs ;o): Is CEP2 maybe not highest priority when choosing -unlimited-? If so, then for the very high core devices suggest the limiter is expanded from '1 default 2-16-unlimited' to '1 default 2-32-unlimited'. Then at least there'd be enough CEP2 queued before any other research starts crow-barring in.