Can't upload trickle-up's, got 4 waiting. The log says that the server is out of disc space.

I also have 1 waiting to upload

OK, I'll inform the project people.

Ha! Beat me to it again, Les. Good on ya. (I have the same issue.)

---

"We have met the enemy and he is us." -- Pogo
Greetings from coastal Washington state, the scenic US Pacific Northwest.

My euro model sent up a trickle just 25 mins ago so maybe the blockage is fixed now. Server status page shows all green

My upload is for HadCM3N. My memory is slippery but I think EU results go to a different server. I'll try to find ...


EDIT: Just forced an upload attempt. Still a problem for HadCM3N returns.

Les' message is unlikely to be seen for hours, as the staff likely snuggles in their warm beds now. So, we wait; in my case, boinc will be denied Internet access in the interim.

---

"We have met the enemy and he is us." -- Pogo
Greetings from coastal Washington state, the scenic US Pacific Northwest.

I've just started uploading an EU zip, and got Scheduler request succeeded, so it could be intermittent. Still, they know now. Or will in 12 hours or so.

PS
IF things are running as they should, files should get moved off the upload servers to backups automatically.

I was just thinking they might be going to different servers as astro posted. Looking through client_state shows:
hadam3p_eu models go to http://cpdn-upload2.oerc.ox.ac.uk/cgi-bin/file_upload_handler
while hadcm3n models go to http://rapid-watch.badc.rl.ac.uk/cpdn_cgi/file_upload_handler.
I would also guess that these may be location-specific to the user.

Still no go, there all in project back off. Mine are all hadcm3n, think I'm up to 6 waiting now, thanks for contacting them Les.

I have a Hadcm file waiting in Transfers too. Flashawk, looking back at your first post in the thread I think you must have meant four waiting files, not trickles. Trickles do get uploaded but they're just 1 or 2 kilobytes in size and I don't think they ever wait in Transfers.

I rather think the upload server for the Hadcm files isn't at Oxford Uni. These models are for a project called RAPIT based at the Uni of Southampton. I think the server must be based at the British Atmospheric Data Centre. The rl in the name suggests to me that BADC is hosted at the Rutherford Laboratory near Didcot. If this is in fact the physical location of the upload server it may mean that Jonathan and Andy don't receive an alarm message when the disc's getting full. That doesn't matter though because J and A will be in regular touch with these people.

If my speculations are accurate I think it will mean that even if this upload server is offline it will not show up on our CPDN Server Status page.

While these files are waiting in Transfers please don't keep pressing the Retry Now button which will have no effect. Files can wait for a long time perfectly safely in Transfers.

---

Cpdn news

What are they called when the model gets to 25%, 50%, 75% etc. They are usually 51 to 52MB in size, those aren't trickle up's? We don't get any credit for those? I have never seen one of these tiny files before, not even a mention of their uploading in the logs. These little trickle up files are going somewhere because there not in my upload queue. I now have over 500MB of hadcm3n files waiting to upload.

Your question indicates a misunderstanding of the way this project works.

To start: You're NOT getting credit for uploading data files. You're getting it for the amount of time that your computer has been running while creating a climate model.

Trickle_up files were originally devised as a means to check on the progress of the early climate models. Back then, all they did was supply a small amount of info that said to the server: "I'm still running, and this is who I am."
A second function was to use each trickle to add some more credit to that computer, for work that it must have done to get to the point of creating that trickle.

Since then, extra use has been made of the trickle_up files, in that a small amount of data is also included in the files. Plus they still have the original 2 functions: "I'm still alive", plus allowing the project to increment the credit.

Trickle_up files are very small, and also do NOT go through the usual BOINC file transfer process. If you want to see them, you'll have to turn off the Network activity in the BOINC manager menu, wait a few hours, and then look at the files under the projects\climateprediction.net project. If you have the files arranged in alphabet order, they'll be towards the bottom. You can use Notepad to open them to see what's inside, but they ARE a form of xml, so be careful.

And they DO get mentioned in the logs. This is a line from mine:

10/04/2013 4:50:07 PM|climateprediction.net|Sending scheduler request: To send trickle-up message. Requesting 0 seconds of work, reporting 0 completed tasks

Lastly, trickle_up files have their own separate server, so they can go through when zip files run into problems.

As for the files that go through the Transfers, these are standard zip files, and contain the bulk of the climate info that your computer has created.

***********

Mo is correct. These files go to another university, and that server is not on "our" list, nor do our project people get warning messages about the status of that server.
The people there have now been informed about their server being full.

.


Thanks very much to Les, Mo, astroWX, Iain Inglis, geophi and many, many more too numerous to mention here ...

all for your excellent and informative posts.

I now have a much better understanding of the way this project works ...

after all these years that I have been here :)


.

Flashawk, our Hadcm models produce a trickle at the end of every model year. This is what tells the server our computer needs more credit for extra crunching. But at the end of every 10 years (25% etc) a file is produced. So each Hadcm model produces 40 trickles and 4 files.

In addition, after every six model days these models checkpoint and that chunk of extra progress is saved. But the checkpoints aren't recorded in our Event log or on the model's web page.

Andy has confirmed that the upload server for these HADcm models is indeed located at the Rutherford (and Appleton) Lab.

Good news - the file I had stuck in Transfers is now uploading.

Thanks to Flashawk and others who reported the problem. I wish all our members would report problems as this is how we sort things out.

---

Cpdn news

OK, thanks mo, I always thought it was the bigger uploads. Anyway, everything's working fine now, all my uploads are gone.

Given past response sequence -- Les is aware and will have already passed the surprise information back to the admins. The admins will likely let Les know they have received his message by sometime on Thursday or Friday and the problem will be temporarily resolved sometime next week <rueful smile of been there done that>

The "paperwork" was completed about 12 hours ago, and apparently "the other people" are onto it.
(We're getting better :) ).

Les, I KNOW the communication/action lag is not on your end <rueful smile, yet again>.

I realize it is a matter of priorities and complexities. Once I notice the upload issue on my systems, I suspended processing on Climate to give other projects some cycles and not add to the Climate backlog.

There is no truth to the rumour that climateprediction.net issues are the cause of the slowdown in global temperature rises over the past few years.

The "paperwork" was completed about 12 hours ago, and apparently "the other people" are onto it.
(We're getting better :) ).

Barry, at the risk of making it worse, I wasn't 'having a go at you'. I was just stating the facts, that this time the problem had been fixed within a couple of hours of first reports, and that I was rather happy about this. I felt that it made a nice change from, as you said the 'several days to fix it' that we usually get.
Even the current FORTRAN problem was dealt with quickly. It's just a shame that Windows people have to find and abort the correct faulty models themselves.

Les, you haven't made it worse at all. I realize you were not 'having a go at me'.

And I should note, that I am appreciative of the quick resolution of this - uploads are now going through and that is very much a good thing.

---