Any way to recover a corrupt pp file?

[pmcjs]

Two of my jobs have corrupt pp files, however, running ls -l shows that they are the same size as all of the other non-corrupt files, so I wonder if the data is in there "somewhere". Running xconv on it returns a segmentation fault. An example of one of them is here on Archer:

/work/n02/n02/pmcjs/aq607a.pm1968sep.pp.corrupt

I've tried xconv, convsh, cumf, pumf and cf-python on these files with no joy.

These jobs happily carried on running all the way to the end of 2014, so the model itself is fine (as are all of the later pp files). I thought I could re-run the offending month by using a start dump from near where the pp files were corrupted (1968 and 1969 in the two jobs), but the runs don't bit compare with the original when I do this and I don't have the Archer resource to do another 90 years of simulation. Because I used ff2pp as well the original ff files were deleted. If you have any other solutions to try I'd be grateful to hear them!

Cheers (and happy Christmas),
Chris

[grenville]

Chris

How were these pp files created - do you have any logs from the conversion?

Grenville

[grenville]

Please set group read permission on your RDF files

[pmcjs]

Hi Grenville,

/work, /nerc and /home now set up for group read access.

Unfortunately I have deleted the logs from the conversion (I set up serial jobs to do this automatically and save the stdout and stderr files but it was a long time ago). I used ff2pp to do the conversion.

Cheers,
Chris

[jeff]

Hi Chris

I had a look at your file and it looks to be mostly zeros, I can read the first 7 fields and the last 4 but everything else is unreadable. I'm not sure what happened and without logs or the original fields files it's hard to say what happened.

Jeff.