-
Notifications
You must be signed in to change notification settings - Fork 3
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Procedure to remove duplicates from analysis breaks code #149
Comments
I am sorry to know it took you so much time. I am going to take a look and see if I can spot anything. On the other hand, it doesn't hurt to by-pass this minor feature. |
I did put code to bypass this ... all worked for me last night. I created IODA files for 4 sync times; I ran jedi.x successfully w/ the first sync time; than jedi crashes w/ a profile check error in the second sync time ... the bug seems to be moving for me. I don't know if's an env thing ... I am still looking into it. |
JJ: don't spend time of this ... I am not quite sure it is an actual bug ... I'm now able to run fine it seems! |
Hi Ricardo, thanks for letting me know. I did try by myself and my test crashed too. But, it was not because of that. I was able to create nc_diag files. It was probably because FVSETUP messed up. It set up lat-long grids for AGCM even though I typed "C360" for the input of AGCM horizontal resolution. Here are my original /archive/u/jjin3/x49atms/run/AGCM.rc.tmpl and my saved FVSETUP input /discover/nobackup/projects/gmao/obsdev/jjin3/SavedInputs/x49atms.input . It needs a fix if you can verify the issue. |
Unfortunately, I spent the entire day today fighting a problem when trying get nc diag files out of the latest 5.30.3 GSI. I pinned the problem to the handling of duplicate removal introduced recently. I am able to get the code to behave as before by resetting the identical_obs array to false right before it gets used to bypass duplicate obs.
I believe the problem is that the way the code is done it leaves entries in the diag arrays filled w/ garbage and the counting done by the link list ends up corrupt.
I am not sure how the binary diags don't seem to get affected - I believe they are actually affected but somehow things get clobbered in a way that the code still more or less works - but I think the obs sequence might be messed up.
Anyway, as is in 5.30.3 the code simply crashes when netcdf diag files are requested; so the code cannot stand as is.
The text was updated successfully, but these errors were encountered: