suggestion for when project finds bad wu's have been sent out

[KSMarksPsych]

Posted by: Date: 1:24 PM 08-25-2006 Just an idea but I think it would be useful to add a piece of code that the projects could use when they discover they have sent out bad WU's or a bad application. The code could allow the project to send a WU with a very short deadline (so it is run ASAP to avoid wasted processing time) aborting the bad WU's they have found or resetting the project if necesary for a bad application. At the moment they only seem to abort on the server end so computers that have already recieved the bad data will waste time processing the bad data unless the users have spotted a message on the servers website telling them what to abort, or at the other end of the scale the users that have read the message might abort all their WU's when only some were bad. Like I said it is just an idea for a future feature.

[KSMarksPsych]

I believe this has been implemented (e.g. killer trickles for CPDN), but will leave it open pending further confirmation.

[MikeMarsUK]

Incidentally there is a trivial bug in Killer Trickle code - if a result is no longer present at the client, you get a 'null pointer' message appearing in the messages tab.

2007-04-24 08:44:01 [Climateprediction.net Beta] [error] handle_trickle_down failed: null pointer

This is completely harmless but tends to result in queries on the helpdesk. As a possible future development a line of text in the messages / website to describe what has happened would be useful in order to reduce the amount of queries.

What needs to be confirmed is that the killer trickle works for non-climate projects as well as normal projects. Is it implemented in the science app or the core client? (I have a feeling it's the science app).

[Nicolas]

There are server-side aborts now, apart from project-specific trickles.