> > You can try the edit "1" to "5" for
> > that _1.zip as Thyme Lawn pointed out, however don't do it on the bit
>
> Carl: I don't understand what you mean by "don't do it on the bit"..?

There may be an "[xml_signed]" section that has the 1000000, don't change that as the "xml_signed" will be validated on the server (for a correct signature), but I have changed the server code to allow 5MB files for each of the _1 to _5.zip files. So it's just the client-side "filter" you need to change, i.e. the "max_nbytes" outside of the [xml_signed] block.

I have identified a small range of runs on the launch version that have this problem; most are kicked out anyway since they had the Fortran namelist error, but a small section (i.e. ones that went to Windows primarily) went out around the launch date. If uk_nick's "trick" of editing the 1000000 to 5000000 works then I will email those people to edit this file. It's a screwup but hopefully not a critical one.

Yay, that worked, Tracy uploaded <a href="http://climateapps2.oucs.ox.ac.uk/cpdnboinc/result.php?field=Temperature&amp;resultid=28012&amp;phase=AT#graph">Result #28012</a> without any error after I edited that 1 to 5, nice catch Thyme Lawn. :-)

See <a href="http://www.climateprediction.net/board/viewtopic.php?p=20943#20943">this post</a> on the phpBB messageboard for specific details of what to look for &amp; edit - Carl has written that he'll email everybody affected.

<a href="http://www.nmvs.dsl.pipex.com/"><img src="http://boinc.mundayweb.com/cpdn/stats.php?userID=6&amp;team=off&amp;trans=off"></a>

&gt; Yay, that worked, Tracy uploaded <a href="http://climateapps2.oucs.ox.ac.uk/cpdnboinc/result.php?field=Temperature&amp;resultid=28012&amp;phase=AT#graph">Result #28012</a> without any error after I edited that 1 to 5,
&gt; nice catch Thyme Lawn. :-)

My pleasure, uk_nick :-)

<a href="http://www.teampicard.net"><img src="http://www.teampicard.net/templates/fisubice/images/phpbb2_logo.jpg"></a><a href="http://climateapps2.oucs.ox.ac.uk/cpdnboinc/team_display.php?teamid=3">Join us here</a>

thanks Thyme &amp; Nick, I've sent out emails to people with this "update" -- hopefully it's the last we've seen of these "surprises!"

&gt;hopefully it's the last we've seen of these "surprises!"
Maybe not Carl! Sorry to throw a spanner in the works, but I'm still running a Beta model sent out on 7th August (01cs_000026741) using hadsm3 version 4.02. I think you were discussing runs that were sent just after 26th Aug.
My client_state.xml file is exactly as Nick described with the 1.zip file set to 1000000.000000 and the rest set at 5000000.000000.
So is there another batch of runs with this problem? And should I edit the client_state.xml file as you described?
Alan

beta runs didn't have the extra code that makes the *_1.zip file so big, so they are "safe." There was just an "overlap" from the CPDN/BOINC launch version, which would make a 1.4MB file with some "old" workunits that still had the 1MB upper limit. The extra code in the launch version (i.e. CPDN/BOINC 4.03+) creates "regional means" (i.e. like your "end of phase" temp &amp; precip chart, but now also by region instead of just global), so it really "ballooned" the zip file. Another lesson in not adding too much at the last minute before a public release!

Thanks Thyme/Nick/Carl.

All eight active runs on my four boxes are "infected". Nice save!

Jim

Just to confirm the fix: Helen's <a href="http://climateapps2.oucs.ox.ac.uk/cpdnboinc/result.php?field=Temperature&amp;resultid=27730&amp;phase=AT#graph">Result #27730</a> uploaded without error. :-)

I've also run Dilly's <a href="http://climateapps2.oucs.ox.ac.uk/cpdnboinc/result.php?field=Temperature&amp;resultid=24494&amp;phase=AT#graph">Result #24494</a> through final result upload from backup and it worked okay - although the server does seem a little confused;

Outcome: Unknown - was 'client error', should be 'success'.
Client state: New - was 'computing', should be 'done'.
Exit status: 0 (0x0) - correct. :-)

but no extra credit this time... ;-)

<a href="http://www.nmvs.dsl.pipex.com/"><img src="http://boinc.mundayweb.com/cpdn/stats.php?userID=6&amp;team=off&amp;trans=off"></a>

Carl:

I'm one of the people who received an email telling me to make this change - and ever since then, I get lots of messages like 'Result xxxxxx exited with zero status but no finished file' and that is immediately followed by notice that it is restarting the calculation.

Is there something that can be done about this?



&gt; &gt; &gt; You can try the edit "1" to "5" for
&gt; &gt; &gt; that _1.zip as Thyme Lawn pointed out, however don't do it on the
&gt; bit
&gt; &gt;
&gt; &gt; Carl: I don't understand what you mean by "don't do it on the bit"..?
&gt;
&gt; There may be an "[xml_signed]" section that has the 1000000, don't change that
&gt; as the "xml_signed" will be validated on the server (for a correct signature),
&gt; but I have changed the server code to allow 5MB files for each of the _1 to
&gt; _5.zip files. So it's just the client-side "filter" you need to change, i.e.
&gt; the "max_nbytes" outside of the [xml_signed] block.
&gt;
&gt; I have identified a small range of runs on the launch version that have this
&gt; problem; most are kicked out anyway since they had the Fortran namelist error,
&gt; but a small section (i.e. ones that went to Windows primarily) went out around
&gt; the launch date. If uk_nick's "trick" of editing the 1000000 to 5000000 works
&gt; then I will email those people to edit this file. It's a screwup but
&gt; hopefully not a critical one.
&gt;
&gt;
&gt;