Message boards : Number crunching : Failed to download
Message board moderation

To post messages, you must log in.

Previous · 1 · 2

AuthorMessage
Profile Tom95134

Send message
Joined: 4 May 07
Posts: 250
Credit: 826,541
RAC: 0
Message 24318 - Posted: 12 Jul 2012, 20:41:19 UTC

"Pumped" the Project and just got 11 new Tasks, all sse2.
ID: 24318 · Report as offensive     Reply Quote
Nuadormrac

Send message
Joined: 26 Sep 05
Posts: 85
Credit: 421,130
RAC: 0
Message 24338 - Posted: 13 Jul 2012, 20:44:28 UTC
Last modified: 13 Jul 2012, 20:50:06 UTC

I'm wondering if this is just all coincidence, of if there's some difference in what tasks the servers are giving out and where. My old laptop, which I'm wrapping up a CPDN task on, has this self same constant download error problem now, last week they didn't all show up this way, after a time before where it cropped up a bit:

http://lhcathomeclassic.cern.ch/sixtrack/results.php?hostid=9957942

The new laptop doesn't show any of these errors though it's connecting as a new box requesting all new tasks:

http://lhcathomeclassic.cern.ch/sixtrack/results.php?hostid=9973530

The operating system is the same, though I didn't do a fresh install of win7 Pro on the new laptop so it's still running win7 home premium (though I have a key for pro, just not sure I want to do the whole format, etc now)... The procs are different core 2 duo vs i7 quad core, but we're looking at a download error, and all those tasks it isn't just my box getting it. They're exceding the quota for total returned tasks or total errors from pretty much everyone... Is this coincidence, are newer boxes not getting these bad WUs, it's enough to make me scratch my head. Maybe a remove, delete project directory, and reattach will work? Has anyone tried? I'm guessing there's still bad tasks that the project is sending out to people left in the queue though, hmm....

The second PC of mine was added yesterday though, vs the other over a year ago, double hmm...
ID: 24338 · Report as offensive     Reply Quote
Profile Igor Zacharov
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist

Send message
Joined: 16 May 11
Posts: 79
Credit: 111,419
RAC: 0
Message 24345 - Posted: 14 Jul 2012, 0:27:11 UTC

When downloading tasks the scheduler is solving a complicated equation
of limits and past execution results. As we had may errors provoked by
the server on our side, hosts became "unreliable" and would not download
tasks, especially which are not new. This will hopefully clear up in the
next few days, as new work is submitted and executed with good results.

We sincerely hope for this on our side.

Igor.
skype id: igor-zacharov
ID: 24345 · Report as offensive     Reply Quote
Nuadormrac

Send message
Joined: 26 Sep 05
Posts: 85
Credit: 421,130
RAC: 0
Message 24354 - Posted: 14 Jul 2012, 7:26:31 UTC - in response to Message 24345.  

But this is the odd thing, the bad tasks seem to be sent to the older client, though I'm not sure how you send the tasks. Some projects (Einstein for instance) tends to send tasks similar to those one had crunched before (so they can re-use some data files and not have to download everything again). I think Rosetta did something similar, though with the way tasks have seemed to disappear on this project, and due to the infrequency of work available I'm not sure such a scheme would help minimize downloads....

But a thought, if for some reason the old client keeps getting sent the download failures or WU rejects (as it tries to meet the 10 quorum), and the new client isn't given that honor :o :lol:

Could be coincidence, or people might try re-attaching after a detach, directory delete, and downloading all new again. But I can't say for certain this could eleviate anyone's download problems as 2 clients, isolated by themselves is rather anectdotal.

Oh, and on one task, I noticed that I ended up validating my own result, as the servers sent me the same task twice. Now I know this comp has a quad core, which is also dual threaded, and given hyper-threading is seen as 8 CPUs. But that was odd seeing my computer come up twice on the WU, and odder yet seeing the run time slightly differed each time (by about a half a minute or whatever), though it validated. Never imagined I'd establish a quorum with myself :o
ID: 24354 · Report as offensive     Reply Quote
Previous · 1 · 2

Message boards : Number crunching : Failed to download


©2024 CERN