Hello mwroggenbuck,
Not that I know of. You can change your cache settings so that you have 0 hours additional cache and connect every 0 days. This means you will only draw a work unit when you have finished a work unit. You will not have any waiting to run. Then you can select only short projects, so even if they are due in 4 days, you have a good chance.

Lawrence

That what Lawrence describes is the first step for very intermittently running computers to not have any jobs queued, albeit, the version 7 client will fetch a new task 5 minutes before a task is projected to finish.

With a client 6.xx [a version before 6.12] you could set the "Switch applications about every..." to some very high number such as 7000 minutes and of course set the Activity menu option to "Run based on preferences". This way the client thinks it would not be allowed to complete a new task in time and report, so it would not fetch the short deadline jobs. This trick stopped working certainly with 6.12. Search the forums for this switch trick or maybe someone can reconfirm which version this still works with. (There is a note somewhere in the Start Here FAQs too IIRC).

--//--

edit: To add, this 7000 minute ''switch'' trick certainly forces any short deadline tasks ahead, meaning if there is a 10 day job processing and a 4 day deadline job arrives, the 10 day job will be paused and the 4 day job will be processed in High Priority. The FAQ mentioned is this one, [so put the reading glasses on ]: https://secure.worldcommunitygrid.org/forums/wcg/viewthread_thread,31893. Will add any version number of client(s) if this is confirmed to work with a 6.10 or above, but it says in the FAQ it only worked on versions prior to 6.10. In that 5.10.45 and 6.2.28 are old faithful and stable clients, in old recommended at WCG.

That's why in the added edit, I put "certainly forces any short deadline tasks ahead" qualification ;-)

It's a high "connect every..." interval of > 4 days that stops the short deadline fetches, but then, you'd not want 4 days of work to be cached on an intermittent running device. From past tests, the BETA's would ignore that IIRC. Could be different with the new server software.

Of course, if you make sure not to report any task within 48 hours, you'd not get them, but then mwroggenbuck noted he sometimes leaves his host running for multiple days and gets these. Certainly the switch setting would with some client versions force these short deadline tasks ahead, so there's least chance of them being on the host with

1: Set swtich 7000 minutes
2: Connect /Min Buf to 0.0 days
3: Set Max Add.Buf to 0.0 days

Not the whole, but certainly half a solution!

--//--

edit: OK, I've asked techs to consider an option who's outcome would work as operating to knock a device permanently into "Not Reliable" state [Reliable being the condition to get short deadline tasks]. That would be at profile level, not host, so the multi device owner who's working with regular and variable running devices would be responsible for associating the appropriate profile to a intermittent host.

Is there any news on this? I still have to abort jobs.

Thanks for the information and suggestion.