Skip to content

Conversation

@xylar
Copy link
Collaborator

@xylar xylar commented Jan 24, 2018

This prevents certain issues with very large NetCDF3 files (e.g. from high resolution MPAS-O output).

Without this fix, an error occurs when running analysis on large MPAS files:

...
netCDF3 NETCDF_64BIT_OFFSET format limits fixed variables to sizes smaller than 2^32 B = 4 GiB ~ 4.2 GB, and record variables to that size per record.

From @czender:

Since the MPAS input is NETCDF_64BIT_OFFSET, that's what ncclimo uses for the output. However, timeMonthly_avg_normalVelocity is > 8 GB, and exceeds the format constraint. (How MPAS wrote the input dataset without error is...unknown to me, perhaps PnetCDF trickery?).

@xylar xylar self-assigned this Jan 24, 2018
@xylar xylar requested a review from milenaveneziani January 24, 2018 23:09
@xylar
Copy link
Collaborator Author

xylar commented Jan 24, 2018

Testing

The QU240 test case on my laptop was successful. I haven't yet tested this branch on other E3SM output.

@xylar xylar added the priority label Jan 24, 2018
@xylar xylar requested a review from vanroekel January 24, 2018 23:12
@xylar
Copy link
Collaborator Author

xylar commented Jan 24, 2018

@milenaveneziani, maybe you can test this on one low-res run and the high-res run that was giving us trouble before? @vanroekel, if you have a run you'd like to try out, that would be great as well.

xylar added 2 commits January 24, 2018 22:15
This prevents certain issues with very large NetCDF3 files (e.g.
from high resolution MPAS-O output).
@xylar xylar force-pushed the nco_force_netcdf4_output branch from 34c6abb to 20cd745 Compare January 25, 2018 05:16
@vanroekel
Copy link
Collaborator

I'm testing this on anvil now.

@milenaveneziani
Copy link
Collaborator

This worked for me on a low-res run on edison. I also tested on the G-case high-res run on titan for which I have a config file that I used before, and I had some trouble with a file that was not found, so nothing to do with this PR.
If @vanroekel's test on the high-res (including MOC) passes, I think we are good. I have tested the -4 option before on the high-res MOC and things had worked at the time.

@czender
Copy link

czender commented Jan 25, 2018

FYI this PR is a workaround for the netCDF CDF5 bug. The fix for that bug is "staged" as a PR for netCDF 4.6.0, which has no release date yet, and will take some months after release before it is installed on all the HPCs. At that point, if anyone cares to, this PR could be reverted.

@xylar
Copy link
Collaborator Author

xylar commented Jan 25, 2018

Thanks @czender. I think we're fine with forcing NetCDF4 output regardless but it will be good to have the bug fixed regardless.

Copy link
Collaborator

@vanroekel vanroekel left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Worked fine on anvil and confirmed climos were netcdf-4 format after mpas-analysis finished.

@xylar
Copy link
Collaborator Author

xylar commented Jan 25, 2018

@vanroekel, thanks for the review.

@milenaveneziani, if you could please approve this PR, I'll merge it. Thanks!

Copy link
Collaborator

@milenaveneziani milenaveneziani left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

sorry, I thought I had approved it already.

@xylar xylar merged commit 11540b1 into MPAS-Dev:develop Jan 26, 2018
@xylar xylar deleted the nco_force_netcdf4_output branch January 26, 2018 00:32
@xylar xylar mentioned this pull request Jan 26, 2018
xylar added a commit that referenced this pull request Jan 26, 2018
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants