climateprediction.net home page
Have my files uploaded?

Have my files uploaded?

Questions and Answers : Windows : Have my files uploaded?
Message board moderation

To post messages, you must log in.

1 · 2 · Next

AuthorMessage
old_user207550

Send message
Joined: 7 Nov 06
Posts: 21
Credit: 43,546
RAC: 0
Message 28612 - Posted: 11 May 2007, 21:50:35 UTC

As my model was approaching 2040, I suspended network activity to prevent the big upload. Following the message in New and Announcements that there is now space on a server, I enabled network activity in early 2041. Messages reported a trickle up, but no sign of the big files I was expecting. What should I do? I can restore if I must.
ID: 28612 · Report as offensive     Reply Quote
Les Bayliss
Volunteer moderator

Send message
Joined: 5 Sep 04
Posts: 7629
Credit: 24,240,330
RAC: 0
Message 28616 - Posted: 11 May 2007, 22:28:06 UTC

The big file would have been visible in the Transfers tab. On my computer it takes about 20-30 minutes to transfer.
There would be / will be be messages about it in the messages tab, on the lines of:
[---] Resuming network activity
[BBC Climate Change Experiment] Sending scheduler request to http://bbc.cpdn.org/bbc_cgi/cgi
[BBC Climate Change Experiment] Reason: To send trickle-up message
[BBC Climate Change Experiment] Note: not requesting new work or reporting results
[BBC Climate Change Experiment] Started upload of hadcm3ohb_ce4k_00878912_0_12.zip
[BBC Climate Change Experiment] Scheduler request to http://bbc.cpdn.org/bbc_cgi/cgi succeeded
[BBC Climate Change Experiment] Finished upload of hadcm3ohb_ce4k_00878912_0_12.zip
[BBC Climate Change Experiment] Throughput 5553 bytes/sec
[---] Suspending network activity - user request


You can see 3 things here:
1) a trickle_up being sent. (And succeeding.)
2) the start of a zip upload.
3) the completion of the zip upload.

Messages are archived in: stdoutdae.txt which is in the BOINC folder.
There is also a separate file: stderrdae.txt with just the error messages.

ID: 28616 · Report as offensive     Reply Quote
old_user207550

Send message
Joined: 7 Nov 06
Posts: 21
Credit: 43,546
RAC: 0
Message 28626 - Posted: 12 May 2007, 8:37:15 UTC - in response to Message 28616.  

The big file would have been visible in the Transfers tab. On my computer it takes about 20-30 minutes to transfer.
There would be / will be be messages about it in the messages tab, on the lines of:
[---] Resuming network activity
[BBC Climate Change Experiment] Sending scheduler request to http://bbc.cpdn.org/bbc_cgi/cgi
[BBC Climate Change Experiment] Reason: To send trickle-up message
[BBC Climate Change Experiment] Note: not requesting new work or reporting results
[BBC Climate Change Experiment] Started upload of hadcm3ohb_ce4k_00878912_0_12.zip
[BBC Climate Change Experiment] Scheduler request to http://bbc.cpdn.org/bbc_cgi/cgi succeeded
[BBC Climate Change Experiment] Finished upload of hadcm3ohb_ce4k_00878912_0_12.zip
[BBC Climate Change Experiment] Throughput 5553 bytes/sec
[---] Suspending network activity - user request


You can see 3 things here:
1) a trickle_up being sent. (And succeeding.)
2) the start of a zip upload.
3) the completion of the zip upload.

Messages are archived in: stdoutdae.txt which is in the BOINC folder.
There is also a separate file: stderrdae.txt with just the error messages.


I can only see the trickle up. Nothing appeared in the Transfers tab. The archived messages are:


To pause/resume tasks hit CTRL-C, to exit hit CTRL-BREAK
2007-05-11 20:08:00 [---] Starting BOINC client version 5.8.16 for windows_intelx86
2007-05-11 20:08:00 [---] log flags: task, file_xfer, sched_ops
2007-05-11 20:08:00 [---] Libraries: libcurl/7.16.0 OpenSSL/0.9.8a zlib/1.2.3
2007-05-11 20:08:00 [---] Data directory: C:\\Program Files\\BOINC
2007-05-11 20:08:00 [---] Processor: 1 GenuineIntel Intel(R) Pentium(R) 4 CPU 2.50GHz [x86 Family 15 Model 2 Stepping 7] [fpu tsc sse sse2 mmx]
2007-05-11 20:08:00 [---] Memory: 1.25 GB physical, 2.98 GB virtual
2007-05-11 20:08:00 [---] Disk: 48.83 GB total, 22.76 GB free
2007-05-11 20:08:01 [climateprediction.net] URL: http://climateprediction.net/; Computer ID: 502689; location: (none); project prefs: default
2007-05-11 20:08:01 [---] General prefs: from climateprediction.net (last modified 2007-05-03 17:22:57)
2007-05-11 20:08:01 [---] Host location: none
2007-05-11 20:08:01 [---] General prefs: using your defaults
2007-05-11 20:08:01 [---] Suspending network activity - user request
2007-05-11 20:08:01 [climateprediction.net] Restarting task hadcm3ohc_1nav_05625214_0 using hadcm3 version 515
2007-05-11 20:13:59 [---] Resuming network activity
2007-05-11 20:13:59 [climateprediction.net] Sending scheduler request: To send trickle-up message
2007-05-11 20:13:59 [climateprediction.net] (not requesting new work or reporting completed tasks)
2007-05-11 20:14:04 [climateprediction.net] Scheduler RPC succeeded [server version 509]
2007-05-11 20:42:12 [---] Exit requested by user

To pause/resume tasks hit CTRL-C, to exit hit CTRL-BREAK
2007-05-11 20:42:27 [---] Starting BOINC client version 5.8.16 for windows_intelx86
2007-05-11 20:42:27 [---] log flags: task, file_xfer, sched_ops
2007-05-11 20:42:27 [---] Libraries: libcurl/7.16.0 OpenSSL/0.9.8a zlib/1.2.3
2007-05-11 20:42:27 [---] Data directory: C:\\Program Files\\BOINC
2007-05-11 20:42:27 [---] Processor: 1 GenuineIntel Intel(R) Pentium(R) 4 CPU 2.50GHz [x86 Family 15 Model 2 Stepping 7] [fpu tsc sse sse2 mmx]
2007-05-11 20:42:27 [---] Memory: 1.25 GB physical, 2.98 GB virtual
2007-05-11 20:42:27 [---] Disk: 48.83 GB total, 22.75 GB free
2007-05-11 20:42:27 [climateprediction.net] URL: http://climateprediction.net/; Computer ID: 502689; location: (none); project prefs: default
2007-05-11 20:42:27 [---] General prefs: from climateprediction.net (last modified 2007-05-03 17:22:57)
2007-05-11 20:42:27 [---] Host location: none
2007-05-11 20:42:27 [---] General prefs: using your defaults
2007-05-11 20:42:28 [climateprediction.net] Restarting task hadcm3ohc_1nav_05625214_0 using hadcm3 version 515
ID: 28626 · Report as offensive     Reply Quote
Les Bayliss
Volunteer moderator

Send message
Joined: 5 Sep 04
Posts: 7629
Credit: 24,240,330
RAC: 0
Message 28631 - Posted: 12 May 2007, 9:23:58 UTC


If you\'ve gone past 2040, and the file wasn\'t uploaded, then it may still be in the Project folder. The size of my 2040 file was 8,465 KB.
There\'s no way to see it on the server after it\'s been sent.

Otherwise, I\'ve no idea what has happened to it.

ID: 28631 · Report as offensive     Reply Quote
old_user207550

Send message
Joined: 7 Nov 06
Posts: 21
Credit: 43,546
RAC: 0
Message 28632 - Posted: 12 May 2007, 10:33:05 UTC - in response to Message 28631.  


If you\'ve gone past 2040, and the file wasn\'t uploaded, then it may still be in the Project folder. The size of my 2040 file was 8,465 KB.
There\'s no way to see it on the server after it\'s been sent.

Otherwise, I\'ve no idea what has happened to it.


I\'m now thinking that something has gone wrong while network access was suspended and I should restore. The ten year file is more important than the restart file as I intend to run to the end. When should the ten year file be created? I will restore to my last backup before that.
ID: 28632 · Report as offensive     Reply Quote
Profile Thyme Lawn
Volunteer moderator

Send message
Joined: 5 Aug 04
Posts: 1283
Credit: 15,824,334
RAC: 0
Message 28636 - Posted: 12 May 2007, 14:59:01 UTC

If you look at your model\'s trickle history you\'ll see that trickles have been recorded for 2039 and 2041, but the 2040 trickle (timestep 3110400) is missing. The upload file is created between there and the next checkpoint.

Looking at your trickle history I\'d guess that you\'d have to be looking for a backup made before 0900 on 10th May. To confirm that you have gone back far enough you should open up the file projects/climatepredition.net/hadcm3ohc_1nav_05625214.xml and check that the date tags give a date earlier than 1st December 2040, which would be this:
   <DAY>1</DAY>
   <MTH>12</MTH>
   <YR>2040</YR>

"The ultimate test of a moral society is the kind of world that it leaves to its children." - Dietrich Bonhoeffer
ID: 28636 · Report as offensive     Reply Quote
old_user207550

Send message
Joined: 7 Nov 06
Posts: 21
Credit: 43,546
RAC: 0
Message 28648 - Posted: 12 May 2007, 19:43:41 UTC - in response to Message 28636.  


Thank you both for the information. Several things there that I did not know. I have restored from a backup taken on the 8th of May 2007 with model date 7/6/2040.

That leaves the question of what happened to my 2040 trickle, the ten year file and the 40 year file. I ran for some days with network activity suspended. Do the trickle files have unique names or is the same one reused? If it is the same file, then my 2041 trickle will have overwritten the 2040 trickle while network activity was suspended. Is the production of the ten and 40 year files dependant upon the contents of the trickle file? This would explain why I lost all three 2040 events.

If this is the case it has implications for everybody who ran with network activity suspended for several model years. I noticed because I was watching for the ten and 40 year files to upload but there could be many models with trickles and 10/40 year files missing.

Andy
ID: 28648 · Report as offensive     Reply Quote
Les Bayliss
Volunteer moderator

Send message
Joined: 5 Sep 04
Posts: 7629
Credit: 24,240,330
RAC: 0
Message 28649 - Posted: 12 May 2007, 20:52:36 UTC

The names of ALL the files are unique. If you look at a trickle_up file, it will have a long string of numbers at the end. These are a time stamp, to destinguish it from all other trickle_ups. This lets you leave the Network disabled for days, and let all of the files accumulate in the project folder, without over-riding each other.

The 40 year data is part of the 10 year zip. A 10 year zip is about 5 Megs, and a 10 year zip at the 40 year point is about 8 Megs.
So there were only 2 files that went missing.
The trickles and the zip files are independant of each other. Both are just a collection of data from what has been calculated and stored since the last time.
The program is quite intelligent in this.


Backups: Here
ID: 28649 · Report as offensive     Reply Quote
Profile mo.v
Volunteer moderator
Avatar

Send message
Joined: 29 Sep 04
Posts: 2363
Credit: 14,611,758
RAC: 0
Message 28663 - Posted: 13 May 2007, 14:48:31 UTC

AndyCheryl asked

Do the trickle files have unique names or is the same one reused?

The first 10-year zip file made at the end of Nov 1930 has an extra _1 added to the model name. The second from 1940 has _2 added, and so on. The last one from 2080 has _16 at the end.
Cpdn news
ID: 28663 · Report as offensive     Reply Quote
old_user207550

Send message
Joined: 7 Nov 06
Posts: 21
Credit: 43,546
RAC: 0
Message 28665 - Posted: 13 May 2007, 18:38:12 UTC - in response to Message 28649.  
Last modified: 13 May 2007, 19:02:15 UTC

Les replied saying
The names of ALL the files are unique.

And
The trickles and the zip files are independant of each other.
...The program is quite intelligent in this.


Thanks for the info. I am still wondering why I lost my 2040 files and what the implications are. I assume from what has been said that trickle files are created regardless of whether network access is available With network access suspended I did not create a ten year file: one did not appear in the Transfers tab and I did not upload one. Is it normal for the program to delay creating the ten year file when network access is suspended and then create it when access becomes available? I wonder whether creation of the trickle file was also inhibited by the same condition within the program? Is the fact that no ten year file was created linked to the fact that there was no trickle file for 2040? Did the program, in effect, look at the trickles to be uploaded (either directly or in some internal results table) when network access became available and decide that there was no ten year file to be created because there was no corresponding trickle file?

Sorry to go on about this but I am concerned that if I had a problem then some others could well have had the same problem. Please could the programmers be made aware of the issue? Conditional statements in programs can become quite complex and unexpected results can arise when an unexpected combination of conditions are present. It’s called a bug :-)

My restored model created and uploaded both files with network access available.

Andy
ID: 28665 · Report as offensive     Reply Quote
Profile mo.v
Volunteer moderator
Avatar

Send message
Joined: 29 Sep 04
Posts: 2363
Credit: 14,611,758
RAC: 0
Message 28666 - Posted: 13 May 2007, 19:02:55 UTC

I\'ll ask one of the Oxford programmers to look at this thread.

The problem of a 10-year file not being created is so rare that I\'m having a hard time trying to remember other instances reported on any of the cpdn forums. On the other hand, many members run their models with network activity suspended most of the time apart from when they specifically enable it and allow their models to trickle. So if suspending network activity prevented the 10-year files being created, we\'d have had hundreds of posts telling us.

The reason for your 10-year file apparently not being created must (I think) be something different.
Cpdn news
ID: 28666 · Report as offensive     Reply Quote
Profile mo.v
Volunteer moderator
Avatar

Send message
Joined: 29 Sep 04
Posts: 2363
Credit: 14,611,758
RAC: 0
Message 28667 - Posted: 13 May 2007, 19:22:28 UTC

Andy, I see that you must have restored this model from a backup last November when it was still pretty new. Have you restored it from backup again recently? Has it passed the 2040 point more than once? If so, the model would only create and send the 10-year file on its first run past that point.

When a backup is restored, it still creates and sends a trickle each year in early December (which the server ignores because it has already received them). But the 10-year files are NOT created again during a restored (repeat) part of a run. They only start being created again when the model passes the point at which it failed and starts crunching new stuff.

Does this explain things?

http://climateapps2.oucs.ox.ac.uk/cpdnboinc/result.php?resultid=6045660
Cpdn news
ID: 28667 · Report as offensive     Reply Quote
old_user207550

Send message
Joined: 7 Nov 06
Posts: 21
Credit: 43,546
RAC: 0
Message 28671 - Posted: 13 May 2007, 22:31:02 UTC - in response to Message 28667.  
Last modified: 13 May 2007, 22:34:04 UTC

Andy, I see that you must have restored this model from a backup last November when it was still pretty new.

Yes, I did. I did not know to close BOINC before I closed the computer.

Have you restored it from backup again recently?

Well, about a couple of months ago. Nowhere near 2040.

Has it passed the 2040 point more than once?

Yes but only as detailed in the above posts. There was no trickle or ten year file the first time so I restored it and it created both after the restore.

If so, the model would only create and send the 10-year file on its first run past that point.

How does it know? Does it ask the server first?


Does this explain things?

http://climateapps2.oucs.ox.ac.uk/cpdnboinc/result.php?resultid=6045660


Er, no. Some ugly looking errors in there. What is it telling me?

Thanks for your help and bringing the thread to the attention of a programmer.

Andy
ID: 28671 · Report as offensive     Reply Quote
Profile mo.v
Volunteer moderator
Avatar

Send message
Joined: 29 Sep 04
Posts: 2363
Credit: 14,611,758
RAC: 0
Message 28677 - Posted: 14 May 2007, 3:13:08 UTC

Hi again Andy

You asked

If so, the model would only create and send the 10-year file on its first run past that point.

How does it know? Does it ask the server first?


No, it doesn\'t ask the server first. The 10-year file is created before the trickle + upload, so the file creation must depend not on instructions from the server but on commands within the model\'s own code. When the commands have operated once, they don\'t operate again.

In the case of your model, because the commands hadn\'t activated on the first run past 2040, they could still be activated second time round.

We will indeed be discussing with the programmers what they think can cause the simultaneous failure to trickle and produce+send the file, and how commonly they think this occurs.

In your case you were vigilant and it\'s because of this, plus the fact that you\'d been backing up, that you were able to force the model to produce the file. I agree that many crunchers would never have noticed and that there may be something that needs to be corrected.

On the graphs there\'s a steep fall visible in a number of models around the 2040 point. I wonder whether there\'s any connection between that and the problem you\'ve raised.

Thank you for your help and patience.

Cpdn news
ID: 28677 · Report as offensive     Reply Quote
Profile mo.v
Volunteer moderator
Avatar

Send message
Joined: 29 Sep 04
Posts: 2363
Credit: 14,611,758
RAC: 0
Message 28694 - Posted: 14 May 2007, 17:47:07 UTC

Milo, who\'s one of the cpdn programmers in Oxford, has read the thread.
Cpdn news
ID: 28694 · Report as offensive     Reply Quote
old_user3

Send message
Joined: 5 Aug 04
Posts: 173
Credit: 1,843,046
RAC: 0
Message 28730 - Posted: 16 May 2007, 10:00:55 UTC

AndyCheryl: Do you still have the backup archive available.
ID: 28730 · Report as offensive     Reply Quote
old_user3

Send message
Joined: 5 Aug 04
Posts: 173
Credit: 1,843,046
RAC: 0
Message 28731 - Posted: 16 May 2007, 10:01:52 UTC

Has anyone else experienced this problem & have a backup archive available.
ID: 28731 · Report as offensive     Reply Quote
old_user207550

Send message
Joined: 7 Nov 06
Posts: 21
Credit: 43,546
RAC: 0
Message 28742 - Posted: 16 May 2007, 19:14:26 UTC - in response to Message 28730.  
Last modified: 16 May 2007, 19:16:24 UTC

AndyCheryl: Do you still have the backup archive available.


Yes, I do. From before and after 2040.

Andy
ID: 28742 · Report as offensive     Reply Quote
old_user3

Send message
Joined: 5 Aug 04
Posts: 173
Credit: 1,843,046
RAC: 0
Message 28780 - Posted: 18 May 2007, 16:44:39 UTC

Can you please upload the backup archive here. Many thanks
ID: 28780 · Report as offensive     Reply Quote
old_user207550

Send message
Joined: 7 Nov 06
Posts: 21
Credit: 43,546
RAC: 0
Message 28797 - Posted: 19 May 2007, 9:37:41 UTC - in response to Message 28780.  

Can you please upload the backup archive here. Many thanks

Yes, I can. You don\'t say whether you want the pre-trickle, post-trickle files or both. I started uploading the pre-trickle directory but it is really slow, it took about half an hour to get to 4% and made the computer so slow as to be unusable. I had to abort the upload. I will start it again tonight and it can run while I sleep.

Please tell Oxford to delete my first attempt. I will upload tonight with a new name.

Which archive do you want? Pre, post or both? If both, it would be easier to put them on CDs and post them if you can give me an address to send them to.

Regards

Andy
ID: 28797 · Report as offensive     Reply Quote
1 · 2 · Next

Questions and Answers : Windows : Have my files uploaded?

©2024 climateprediction.net