climateprediction.net home page

The world's largest climate forecasting experiment for the 21st century.

Posts by [B^S] darkclown

1) Message boards : Number crunching : Report deadline now reporting a date of 1901 (Message 25620)
Posted 2680 days ago by [B^S] darkclown
It only matters if you also run other, short projects, which DO enforce the deadline.


I\'m on a Core 2 Duo machine, and only run one other project at any given time, so CPDN has its own core 99%+ of the time.

Guess I\'ll just let it be for now.
2) Message boards : Number crunching : Report deadline now reporting a date of 1901 (Message 25616)
Posted 2680 days ago by [B^S] darkclown
The cmos backup battery is probably flat, and the computer at one point reverted to the builtin default of 1901.
You need to find the time in client_state.xml and change it.
AFTER MAKING A BACKUP!

\"Time\" is mentioned in several places, and I think that you need to change it for each project.

How to do this was described somewhere by Thyme Lawn on the php board.



Knowing that CPDN ignores Report Deadline, is there any reason I need to fix this, other than not liking to look at the 1901 deadline?

I doubt it\'s my CMOS battery; the date was reset to 2002, not earlier.

I did a search on Time at the old php boards; nothing came up.
3) Message boards : Number crunching : Report deadline now reporting a date of 1901 (Message 25610)
Posted 2680 days ago by [B^S] darkclown
System crashed earlier while BOINC was running, processing a CPDN WU. Upon recovery, I found my system clock set to 2002, and BOINC now shows my CPDN WU with a Report Deadline of 12/13/1901. I know CPDN ignores the deadline, but is there a way to fix this, as it throws my system into EDF mode?
4) Message boards : Number crunching : no trickles? (Message 25316)
Posted 2697 days ago by [B^S] darkclown
Yes, there\'s been another problem with the trickle server for the last couple of days.
If it\'s not one thing, it\'s another.

So now we all have to wait until Monday morning Oxford time for the server stats, and Tuesday sometime for the external stats sites.



That would explain why my stats have seemed static for the past few days, when I have 2 machines sending up trickles (and both passed the 1960 mark this weekend!)
5) Message boards : Number crunching : no trickles? (Message 25241)
Posted 2701 days ago by [B^S] darkclown
Could it have queued up the multiple trickles for 03:11, and sent it at the same time as the 16:37, and only logged one message?

Yes, that\'s how it works.
If you wait until the trickles get listed on the trickle server, you\'ll be able to see all of the ones that got sent in a bunch. (This program runs twice a day, I think.)
The times for each will generally be the same, with perhaps a difference in the seconds for some. And the \"timesteps\" will be out of order. I\'ve noticed when I\'ve \"batched\" a large number, that they are often in reverse order.



Ok, thanks for the clarification. :)
6) Message boards : Number crunching : no trickles? (Message 25239)
Posted 2702 days ago by [B^S] darkclown
Your firewall settings may have been changed by the last lot of Windows updates.

If a wu \'gets broken\', it will send a message to the server to say so. In which case there will be a message to this effect in the Messages tab.
(And in stdoutdae.txt, which is a continuous archive, and is in the BOINC folder.)
You can also tell if it\'s crashed by the Status of the model in the Tasks tab.

If it\'s a firewall problem, there will be an accumlation of files in the folder with the model\'s name. These will start with: trickle_up_, followed by the model name, followed by a string of numbers, which is a time stamp.



I don\'t use the Windows Firewall, and none of my other projects running have issues communicating with their respective sites. There are no \"trickle*\" files in my BOINC folders, and nothing in the log file. The model is still running, but based on its current TS vs. the last reported TS, I should have sent several trickles by now.

EDIT: Ok, on one of my WUs, http://climateapps2.oucs.ox.ac.uk/cpdnboinc/result.php?resultid=5809391 , I see that it registers multiple trickles at 11/22 03:11, and then one at 11/22 16:37. Looking through my logs, I can find the trickle message corresponding with 16:37, but not the ones for 03:11. Could it have queued up the multiple trickles for 03:11, and sent it at the same time as the 16:37, and only logged one message?
7) Message boards : Number crunching : no trickles? (Message 25236)
Posted 2702 days ago by [B^S] darkclown
I\'ve noticed that for the past few days, my WUs are not sending trickles up. I can\'t figure why, unless this means the WUs are broken and not worth processing anylonger, which I hope isn\'t the case.
8) Questions and Answers : Windows : No Credits After 15 Hours CPU Time (Message 24596)
Posted 2748 days ago by [B^S] darkclown
Thank you, this answers my question (I had thought trickles were daily)I\'ve got as far as May 1921.


15 hours might not be enough, depending what else you might be crunching, the speed of your computer, etc. Trickles are sent back every project year. I have one machine that sends a trickle back every about 30-40 hours.

I believe the trickles are created every 01-Dec of each project year.

Keep a watch, the credits will come.




Every 0.625% from what I recall.
9) Questions and Answers : Windows : Backups - How to, and why (Message 24511)
Posted 2753 days ago by [B^S] darkclown
People who run multi-project need to read my posts on this subject on these boards, or this page in the BOINC Wiki.


Ah, thanks. Bit of a pain, but eh, nothing is easy in life :)
10) Questions and Answers : Windows : Backups - How to, and why (Message 24506)
Posted 2753 days ago by [B^S] darkclown
If you\'re backing up and restoring your entire BOINC folder, what will that do to other active projects when you restore them - take them back as well? This could be a major headache, esp. if some of that work had been submitted, etc.


Next 10 posts



Copyright © 2002-2014 climateprediction.net