Hi

I have a zipfile that is failing every upload attempt.
Checking back through the logs I find that it has the same message every time.
The first upload attempt was in February.

28/03/2013 7:02:40 PM | climateprediction.net | Started upload of hadcm3ilse_l03m_1980_100_07528236_1_3.zip
28/03/2013 7:02:47 PM | climateprediction.net | [error] Error reported by file upload server: can't open file
28/03/2013 7:02:47 PM | climateprediction.net | Temporarily failed upload of hadcm3ilse_l03m_1980_100_07528236_1_3.zip: transient upload error

Trickles are going through OK.

Can anything be done to fix this?

Bruce

Hi Boss

I have had three _4 Hadcm files upload without problems today so there isn't anything wrong with the upload server. This file must have been stuck in the Transfers tab since 23 Feb. Yes, I can see the trickles all listed correctly on the model's web page so I see nothing wrong with the computer's web connection.

I can't think of anything we can do to make the file upload. As the model must be within a few model years of completion if I were you I'd just wait to see whether file _4 uploads correctly. Don't keep using the Retry now button in the Transfers tab for file _3 because AFAIK BOINC only allows 100 upload attempts for individual files. File _3 won't time out in Transfers until IIRC three months after the first upload attempt.

If _4 uploads OK but _3 stubbornly refuses you will have sent up the final data for this 40-year period plus the handover data for the next computer to compute the next 40 years of the time series that the model belongs to.

Has this computer been rebooted since the upload problem began?

---

Cpdn news

If you haven't exited boinc and restarted it, I would do that in an attempt to get it unstuck. It may or may not work. While you're at it, you might as well reboot as mo suggests.

Hi Mo.v,

As the model must be within a few model years of completion if I were you I'd just wait to see whether file _4 uploads correctly.

This model is at ~37% complete.

Don't keep using the Retry now button in the Transfers tab for file _3 because AFAIK BOINC only allows 100 upload attempts for individual files. File _3 won't time out in Transfers until IIRC three months after the first upload attempt.

I only enable network when a trickle is ready to be sent.

Has this computer been rebooted since the upload problem began?

The original machine the model was running on died mid march. (An AMD K6-2 which only completed 30% in 15 months run time)
The logs say this macine gave up trying to upload the file.

A couple of days ago I have transferred the model to a temporary VMware machine on an Intel-i5 to finish it. It is running MUCH faster (~30x) and may complete around deadline.
This machine is re-attempting to upload the file, but the same error occurs.


I will wait and watch.

PS: The Zip file has been saved to a Backup folder, so that it could be manually sent to the scientists if necessary.[/quote][/i]

Transferring a model to a much faster machine causes it to run into a BOINC limit.

I forget the details, which seem to be on our php server, but it's something to do with the amount of time that was allocated when you started the model.
I think that it's possible to (carefully) edit some values in client_state.xml, but ...

I don't think this particular model is likely to run into the maximum time exceeded error (which I think is now called maximum elapsed time exceeded) because it was transferred after 75% completion. This should allow the second faster machine to complete the model even if its computing speed is three times faster. But I would take a backup each week of the complete contents of the BOINC Data directory because if by chance the model did hit the maximum computing time allowed, this is what you'd have to restore.

If the speed difference is greater than that you could run into the error. If this is the case let us know and we'll investigate which file needs to be edited to avoid hitting the wall.

The last file should upload fine. I transferred a model from a computer that was dying and the files produced after the transfer uploaded without problems. That was a BBC model years ago; I can't think of any new developments that would make the situation now. (That's on CPDN. Most projects don't allow tasks to be moved to a different computer.)

---

Cpdn news

I finally got some time to post update.

The model has now completed processing and is is "uploading" status.

I now have zip files 3, 4, 5, 6, 7, 8, 9 and 10 all reporting the same problem.


climateprediction.net | [error] Error reported by file upload server: can't open file



There is nothing I can do, so am just waiting and allowing network for a short time once a day to see if the problem has been fixed.


Bruce

I have the same thing on the laat model completed. The last 5 trickels are waiting to upload now. Mine hang up at 100% and I have credit for them already so it looks like the upload server is just not letting me system know that it has them.

I have been traveling and the trickels were completed in two states in the US far from where I live. Had this happen before and when I got home they reported.

Also if there is a problem with the upload server it may not show up to many people when most are out of work for this project. Hope there is more work before to long.

Cheers
Ray

I have credit for them already so it looks like the upload server is just not letting me system know that it has them.

The credit is granted on the trickles so you get the credit even if the zips haven't gone. Best thing to do is suspend network activity till you see a message here saying that the problems with whichever server it is are resolved.

Woohoo

Files began uploading successfully today.

All uploaded, reporting completed and shows successful on server.

:) :) :)