Would it be worth putting a note to the effect of \"read Thyme Lawn\'s entry in the Discussion Forum re: resetting the upload time for failing uploads\" in the main news section? I\'ve been happily continuing to work on a project and waiting for the previous one to go up once the server problems are sorted, unaware that there would be a timeout problem, and have just tonight lost all the uploads with messages of:

\"climateprediction.net - 2005-09-05 22:11:46 - Giving up on upload of 0rgf_100055963_0_4.zip: too much elapsed time\"

If I had known this would be a problem and that there was a solution on the discussion boards, I would have followed the solution and thus saved the files and their data. The message \"this isn\'t a problem because it will all sort itself out when the servers are fixed\", which is what the \"latest news\" section says, led me to believe that I could just let it carry on and it would all be back up soon. It occurs to me that many other people could be about to lose their results data in the same way.

I check the \"latest news\" regularly, but don\'t have time to read through all the stuff on the forums on all the projects I run under BOINC, so something this important will only get to me if it\'s referred to in the news.

Hope this is helpful.

It's something that I have feared for a while.
But then there will be those who NEVER read anything on the site until it's too late.

If you have a backup from the time of model completion, you could save the lot to dvd, which is the only reliable solution that I can think of. Then, someone, sometime, may come up with a "cure".

As I've posted elsewhere, I've tried Th's edit idea, and it didn't work for me.
And, as I posted <a href="http://www.climateprediction.net/board/viewtopic.php?t=3152&amp;start=30"> here,</a> there may be a way out of it.
Just wish I knew when Bern will be back. If it's been this long, it could be a LOT longer.
Which means there will be lots more tales of woe. (WOE!)

9/5/2005 11:30:40 PM|climateprediction.net|Temporarily failed upload of 3v5p_100202159_1_4.zip: -106
9/5/2005 11:30:40 PM|climateprediction.net|Giving up on upload of 3v5p_100202159_1_4.zip: too much elapsed time
9/5/2005 11:30:40 PM|climateprediction.net|Started upload of 3v5p_100202159_1_5.zip
9/5/2005 11:31:01 PM|climateprediction.net|Temporarily failed upload of 3v5p_100202159_1_5.zip: -106
9/5/2005 11:31:01 PM|climateprediction.net|Giving up on upload of 3v5p_100202159_1_5.zip: too much elapsed time
9/5/2005 11:31:02 PM|climateprediction.net|Sending scheduler request to http://climateapps2.oucs.ox.ac.uk/cpdnboinc_cgi/cgi
9/5/2005 11:31:02 PM|climateprediction.net|Requesting 0 seconds of work, returning 1 results
9/5/2005 11:31:03 PM|climateprediction.net|Scheduler request to http://climateapps2.oucs.ox.ac.uk/cpdnboinc_cgi/cgi succeeded

I assume from Les Bayliss's post that there is no way at present to upload these results after you get the "giving up..." message, even though the results are still on the disk. Please let me know if this is the case. Thanks.

I think the people who are managing this project are responsible for prominently displaying on the climateprediction web site that there is something terribly wrong and instructions for saving results (if possible). There's no need to force participants to rummage through the discussion forums. If you look at the web site, you'd think nothing at all is wrong, when the whole project is actually locked up.

New message in news <a href="http://www.climateprediction.net/newsb.php?id=0"> here</a> about Bern problem, and from Carl <a href="http://www.climateprediction.net/board/viewtopic.php?t=3152&amp;start=30"> here.</a>

&gt; It's something that I have feared for a while.
&gt; But then there will be those who NEVER read anything on the site until it's
&gt; too late.
&gt;
&gt; If you have a backup from the time of model completion, you could save the lot
&gt; to dvd, which is the only reliable solution that I can think of. Then,
&gt; someone, sometime, may come up with a "cure".
&gt;
&gt; As I've posted elsewhere, I've tried Th's edit idea, and it didn't work for
&gt; me.
&gt; And, as I posted <a> href="http://www.climateprediction.net/board/viewtopic.php?t=3152&amp;start=30"&gt;
&gt; here,</a> there may be a way out of it.
&gt; Just wish I knew when Bern will be back. If it's been this long, it could be a
&gt; LOT longer.
&gt; Which means there will be lots more tales of woe. (WOE!)
&gt;
&gt;
&gt;I lost 2 sets of data (home and work) because I thought the problem was just due to the connection. I then had the timeout message at home. Too late. When o got to work the same thing. I am frustrated about this as the data takes so long to crunch and don't have time so go trawling through stuff. I'm just an ordinary joe trying to do my bit.

&gt; I am frustrated about this as the data takes so long to crunch and don't have time so go trawling through stuff. I'm just an ordinary joe trying to do my bit.

The possibility that a server site might be put out of action for a prolonged period by some disaster such as the Bern flash flood evidently didn't occur to the BOINC designers. To be fair to them, they aren't he only ones whose disaster planning has been shown to be inadequate recently. It's a salutary reminder to all of us that things that may seem very simple to fix on paper may not be if those responsible locally are coping with other priorities (which I imagine is what happened at Bern), lack electricity or communications, can't get to the site, or whatever.

BOINC was designed to be an 'install and forget it' application, but it may be unrealistic to assume that it will ever cope with everything.

I believe the redirection from the Bern server is working now? I apologize for anyone who has missed an upload due to the long downtime of this server. We always envisioned that in the event of a catastrophe that we would be able to get a server back online within the two week "window" that BOINC allows before giving up. The problem with the upload servers on CPDN is that due to lack of resources they are donated, and we can fall to the "back of the queue" at an institution.

BOINC now has the facility to "round robin" a list of upload servers -- that has been in place after we had generated hundreds of thousands of workunits -- but we will probably regenerate them with all of our upload servers listed so if one server is down it will just move on to the next one.