Advanced search

Message boards : Number crunching : All "CRYPTICSCOUT_pocket_discovery" cancelled by server

Author Message
Erich56
Send message
Joined: 1 Jan 15
Posts: 944
Credit: 3,683,395,665
RAC: 853,259
Level
Arg
Scientific publications
watwatwatwatwatwatwatwatwat
Message 58122 - Posted: 16 Dec 2021 | 6:48:50 UTC

yesterday, one of my machines received several WUs named "CRYPTICSCOUT_pocket_discovery..."
Since an "ADRIA bandit" WU was still in process, the new ones were in waiting position.

This morning, I noticed that all these "Crypticscout" WUs were cancelled by the server at different time during last night.
Why so ?

Ian&Steve C.
Avatar
Send message
Joined: 21 Feb 20
Posts: 744
Credit: 4,943,693,494
RAC: 515,366
Level
Arg
Scientific publications
wat
Message 58126 - Posted: 16 Dec 2021 | 13:47:18 UTC - in response to Message 58122.

these seem to be distributed in pairs, but they only need one result. my guess is that they want to ensure they get the results fast. if they get one result and you haven't even started processing yet, there's no point in having you waste computation on a result they already have. so they cancel it. if you would have been in progress, it wouldn't have been cancelled.
____________

Bedrich Hajek
Send message
Joined: 28 Mar 09
Posts: 411
Credit: 6,063,938,459
RAC: 14
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 58128 - Posted: 16 Dec 2021 | 14:26:21 UTC - in response to Message 58126.

these seem to be distributed in pairs, but they only need one result. my guess is that they want to ensure they get the results fast. if they get one result and you haven't even started processing yet, there's no point in having you waste computation on a result they already have. so they cancel it. if you would have been in progress, it wouldn't have been cancelled.


That is correct, and I think it would be a good idea doing this on all the WUs, maybe waving it off when they have too many WUs.




Erich56
Send message
Joined: 1 Jan 15
Posts: 944
Credit: 3,683,395,665
RAC: 853,259
Level
Arg
Scientific publications
watwatwatwatwatwatwatwatwat
Message 58129 - Posted: 16 Dec 2021 | 15:02:04 UTC - in response to Message 58126.

these seem to be distributed in pairs, but they only need one result. my guess is that they want to ensure they get the results fast. if they get one result and you haven't even started processing yet, there's no point in having you waste computation on a result they already have. so they cancel it. if you would have been in progress, it wouldn't have been cancelled.

thanks for the explanation, good to know.
So there is no use in having them sit in the queue for too long time.
Had I know this before, I would have interrupted the Adria task and given priority to the Crypticscout task.
Too bad that no one from the team explains this beforehand :-(

Ian&Steve C.
Avatar
Send message
Joined: 21 Feb 20
Posts: 744
Credit: 4,943,693,494
RAC: 515,366
Level
Arg
Scientific publications
wat
Message 58130 - Posted: 16 Dec 2021 | 15:55:43 UTC - in response to Message 58129.

these seem to be distributed in pairs, but they only need one result. my guess is that they want to ensure they get the results fast. if they get one result and you haven't even started processing yet, there's no point in having you waste computation on a result they already have. so they cancel it. if you would have been in progress, it wouldn't have been cancelled.

thanks for the explanation, good to know.
So there is no use in having them sit in the queue for too long time.
Had I know this before, I would have interrupted the Adria task and given priority to the Crypticscout task.
Too bad that no one from the team explains this beforehand :-(


I wouldn't worry about it. you don't get more credit per unit time for crypticscout vs adria or anything like that so there's no reason to interrupt an Adria to prioritize CS. and you didn't waste any time on it for no credit since the task never even started on your system. no harm, no foul.
____________

Erich56
Send message
Joined: 1 Jan 15
Posts: 944
Credit: 3,683,395,665
RAC: 853,259
Level
Arg
Scientific publications
watwatwatwatwatwatwatwatwat
Message 58131 - Posted: 16 Dec 2021 | 16:53:36 UTC - in response to Message 58130.

I wouldn't worry about it. you don't get more credit per unit time for crypticscout vs adria or anything like that so there's no reason to interrupt an Adria to prioritize CS. and you didn't waste any time on it for no credit since the task never even started on your system. no harm, no foul.

The situation was a different one:
I had 1 Adria running and 3 CS in the queue. So with the current situation that new WUs become availble scarcely, I would have had 3 more to crunch once the Adria got finished. So, Adria got finished, and no other ones left :-(

Ian&Steve C.
Avatar
Send message
Joined: 21 Feb 20
Posts: 744
Credit: 4,943,693,494
RAC: 515,366
Level
Arg
Scientific publications
wat
Message 58133 - Posted: 16 Dec 2021 | 18:06:55 UTC - in response to Message 58131.

but they were crunched already by someone else. so why waste the time on it? similarly, it would be the same if you completed first. why should someone else waste the time on it?

unless you want to crunch tasks unnecessarily just to ensure you get more points?
____________

Erich56
Send message
Joined: 1 Jan 15
Posts: 944
Credit: 3,683,395,665
RAC: 853,259
Level
Arg
Scientific publications
watwatwatwatwatwatwatwatwat
Message 58139 - Posted: 16 Dec 2021 | 19:56:25 UTC - in response to Message 58133.

but they were crunched already by someone else. so why waste the time on it?

yeah, you are right !

Post to thread

Message boards : Number crunching : All "CRYPTICSCOUT_pocket_discovery" cancelled by server

//