WU cancelled
15 Oct 2016 9:58:24 UTC
Topic 201960
(moderation:
Hi Staff,
why are so many WU's are cancelled by the Project (i.e.Workunit 257481933 and many other 43 between 12.-14.10.2016)?
When it is necessary to do so why wouldn't deleted such WU's by the Client Side so that they couldn't waste Time and Power.
I've to kill all old WU's in my caches self to get new WU's wich run fine so far as before. If some WU's are "bad" and couldn't delete automatically by the Project it would only be fair to make a Note from the Project Team.
BTW: I would prefer outside the News Section on the Website a closed Forum branch there "only" the Project Staff can make postings about Announcements and Notes of actual Problems and Plans of e.g. new Apps and WU's "before" they are going to start.
BR and Thx for the good Jobs so far
Greetings from the North
I am NOT Staff but will
)
I am NOT Staff but will try....sometimes they are resends that auto occur when someone misses their deadline but then returns it before we start crunching it, if we did finish the unit we would get no credits for it, so they just cancel it to save us the trouble. Other times they are units that have 'issues' of one kind or another and they just cancel batches of them and send out new ones. I'm sure there are other reasons too, but I can't think of any at the moment.
There already is a section in
)
There already is a section in the forums where we post stuff like this: Technical News
The workunit you mentioned is part of a bad beam from Arecibo where noisy data slipped through our pre-processing. That happens from time to time and we usually cancel all jobs that belong to that particular beam. The thread you may want to monitor is: Canceled "bad" BRP beams
Canceling jobs on the server side does not automatically cancel tasks on the Client side. There is a mechanism that should cancel unstarted tasks on the Client but it I'm not sure if this is available in the server code we currently use on Einstein@Home. I'll have a look into this after the GW run has settled down.
Edit: I should have followed my own link above. It turns out that the send_result_abort feature has some performance problems and so it was deactivated on Einstein@Home:
Ugh thats why my Pi2 had some
)
Ugh thats why my Pi2 had some invalided tasks. 77-80k seconds on each. Quorums were not canceled though. I got three on the 13th-14th while the canceled post was on the 12th.
https://einsteinathome.org/host/12441987/tasks/invalid
I've had other projects cancel WUs of mine where I was the 2nd wingman but the 1st wingman completed it right before their deadline and mine was aborted by the server. Is that the feature that was disabled?
mmonnin wrote:I've had other
)
Yes that's the feature.