I\'m running Model hadsm3fub_0347_005916009_5 (UK Met Office HadSM3 Slab Model 5.06),
Boinc 5.10.13, WinXP prof. SP2, Pentium 4.

When the model finished, it told me that output files hadsm3fub_0347_005916009_5_1.zip,
hadsm3fub_0347_005916009_5_2.zip and hadsm3fub_0347_005916009_5_3.zip were not available
and aborted with error.
Since my network connection is disabled (due to previous boinc errors) nothing was sent.
So I went back to the latest backup and restarted it. But the result was the same.
In the project dir those files were missing indeed, but there were three other files named
hadsm3fub_0347_005916009_1.zip, hadsm3fub_0347_005916009_2.zip, and hadsm3fub_0347_005916009_3.zip.
In client_state.xml I found in the result section
<file_ref>
<file_name>hadsm3fub_0347_005916009_5_1.zip</file_name>
<open_name>cpdnout1.zip</open_name>
</file_ref>
<file_ref>
<file_name>hadsm3fub_0347_005916009_5_2.zip</file_name>
<open_name>cpdnout2.zip</open_name>
</file_ref>
<file_ref>
<file_name>hadsm3fub_0347_005916009_5_3.zip</file_name>
<open_name>cpdnout3.zip</open_name>
</file_ref>
but Boinc had created those files with \'_5\' missing. Why?
And is there a possibility to finish the model successfully (the result
files exist, but have a different name!)?
My last valid backup is with about one day to go.
Shall I edit file client_state.xml or any other file?
Or shall I abort the model?


Sorry if this had been addressed before - I couldn\'t find it...

Matthias

This is the link to that specific model:

http://climateapps2.oucs.ox.ac.uk/cpdnboinc/result.php?resultid=7063246


The phase one and phase two data has been successfully uploaded to the server, it is just phase three which is in trouble. If it has been restored from a backup, but then crashes at the same point, then I think you\'ve done all you can try.


hadsm3fub_0347_005916009_1.zip is a different upload to hadsm3fub_0347_005916009_5_1.zip - the _1 is what is uploaded after phase one, but _5_1 is presumably uploaded after the end of phase 3.

It is important to realise that the reason the model crashed is (currently) unknown, and not to do with the zip file uploads - if a model crashes before the end, then not all the zips will exist yet, and you\'ll get messages like the ones you received.

IF you want, you could restore the backup again and see if it reaches the end on the third try, but personally I usually give up after the first restore!

---

I'm a volunteer and my views are my own.
News and Announcements and FAQ

The phase one and phase two data has been successfully uploaded to the server, it is just phase three which is in trouble.

I managed to complete and upload the model by manipulating the client_state.xml file. But I find the task details display rather irritating in this case because it suggests that data for phases one and two has been uploaded (P1 and P2 were visible and active) although no upload took place before.
So is it an error or a feature that the visibility of P1/P2 depends on the corresponding trickle and not on the successful upload of the phase 1/2 result file? And if it is clicked - why is there any data visible without download? Does it depend on the trickle information only?

The model in question is a \"slab\" model. (All of the current slabs are labelled fub near the start of their name.) The data for these models is ALL in the zip files; their trickles are only to tell the server that the model is still being run.

If you look at the list of trickles on the Trickles page for that model, you\'ll notice 2 things:

1) All of the trickles for all 3 phases are there. (There\'s 24 in each phase.)

2) The last trickle in phase 1 was uploaded on the 03 Jan 2008 20:40:42, one of a batch of 13. This is when the phase 1 zip would also have been uploaded.
The last trickle in phase 1 was uploaded on the 29 Feb 2008 13:21:12, one of a batch of 8. This is when the phase 2 zip would also have been uploaded.

So there isn\'t any problem that I can see.

---

Backups: Here

Yes, all trickles have been sent, but due to the file name mismatch (reason unknown) all zip files for phase 1-3 were uploaded on 10 April. But with P1/P2 I could see Temperature and Precipitation before that date although no zip files were uploaded at all. This looks a bit strange to me...
You can take a look at the UK Met Office HadSM3 Slab Model hadsm3fub_016f_005917938:

http://climateapps2.oucs.ox.ac.uk/cpdnboinc/result.php?resultid=7082570

It is currently in phase 2, all trickles have been received and P1 shows some data, but the phase 1 zip file hadsm3fub_016f_005917938_5_1.zip has not been uploaded yet (blocked by firewall). So I wonder what data is shown if I press P1...
And the mere visibility of P1/P2 is no evidence for the successful upload of the corresponding zip file.

OK. I\'ve just had a report back from a remote area of my memory.

When the slabs were revived for this new project, they were changed a bit.
The final trickle for each phase now also has the data needed to create the graphs shown to crunchers. Which is where your graphs are coming from.
The full data used by the researchers is still in the zips.

I\'ve no idea why your zips (your first post), didn\'t have the correct file names. What was in client_state.xml were the correct names.

---

Backups: Here