Skip to content
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.

Sign up for GitHub

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

Jump to bottom

PIO warning in acme_developer test RP_Ln9.ne30_ne30.FC5.cori-haswell_intel.cam-outfrq9s #1455

Closed
ndkeen opened this issue Apr 25, 2017 · 4 comments
Closed

PIO warning in acme_developer test RP_Ln9.ne30_ne30.FC5.cori-haswell_intel.cam-outfrq9s #1455

ndkeen opened this issue Apr 25, 2017 · 4 comments
Assignees
@jayeshkrishna
Labels
Cori Machine Files

Comments

@ndkeen
Copy link
Contributor

ndkeen commented Apr 25, 2017 •
edited
Loading

I'm seeing a PIO warning (error?) with a test in acme_developer and not sure if it should be there or not.

On cori-haswell with Intel compiler, with master. I have been changing the number of MPI's, but it looks like every attempt has these messages:

pio_rearranger value, 2 , not supported - using PIO_REARR_BOX

/global/cscratch1/sd/ndk/acme_scratch/nc-haswell-pelayouts/ERP_Ln9.ne30_ne30.FC5.cori-haswell_intel.cam-outfrq9s.20170424_161759_0ul4p2

Grepping in all acme.log* from previous acme_developer runs:
I can verify that none of my edison acme_developer tests have this warning.
I can verify that no other cori-haswell or cori-knl test has this warning.
I can verify that I see this warning from previous runs of acme_developer on haswell at least 30 times dating back to Jan -- some of these even passed (ie may not be related to test failing/passing)
I can verify that at least some (not sure if it's all) of these same tests on cori-knl also have this warning.
@jayeshkrishna
Copy link
Contributor

@ndkeen : Can you check the value of PIO_VERSION in env_build.xml (it should be 1)?

agsalin pushed a commit that referenced this issue May 2, 2017
@jedwards4b
Merge pull request #1455 from jtruesdal/newmaps
5619a82 
add new CESM gx1v7 runoff maps for use with estuary parameterization.
Updated CESM mapping files that deal with gx1v7 runoff when using EBM (default for gx1v7). These new files take a merged approach which 1) maps liquid runoff from the nearest coastal point in open ocean where EBM is active and 2) smoothly distributes liquid runoff over a spreading radius in marginal seas where EBM is currently not active. Ice runoff is always spread. gx3v7 maps for use with EBM are in the repository but EBM is off by default for resolution other than 1 degree. Where EBM is not actively mixing we use the spreading maps for both liquid and ice.

Test suite: cime regression and one off tests to make sure the new datasets are being run for the gx1v7 but not for other resolutions.
Test baseline:
Test namelist changes:
Test status: climate changing. Effects analysed and validated with science runs.

Fixes

User interface changes?:

Code review:jedwards
jgfouca pushed a commit that referenced this issue Jun 2, 2017
@jedwards4b
Merge pull request #1472 from mnlevy1981/mapping/runoff_to_gx1v6
baaa2cc 
Add new runoff -> gx1v6 maps
#1455 updated runoff maps for gx1v7 and gx3v7 grids, but some users may still be dependent on the gx1v6 grid and the old maps will trigger aborts if the Estuary Box Model is turned on in POP on the gx1v6 grid. This pull request provides new maps for that resolution to avoid errors, and also the namelists needed to generate these new maps with our runoff tool.

Test suite: Ran a B1850 compset out of cesm2_0_alpha06i, but using a branch that errors out if the EBM does not like where the runoff is coming in (a feature coming to the pop trunk soon); verified that the old default maps triggered this error, and that the new maps do not.
Test baseline: N/A
Test namelist changes: N/A
Test status: changes answers when running with a gx1v6 resolution because runoff is mapped differently

Fixes: N/A

User interface changes: N/A

Code review:edwards trusdale
@ndkeen
Copy link
Contributor Author

ndkeen commented Jul 27, 2017

That particular test is no longer in the tests, but we do have ne4 versions which still have the same warning. I did check that indeed the value of PIO_VERSION is 1.

After the upgrade on edison, I ran the tests and here is the case dir where I see the warning:

/global/cscratch1/sd/ndk/acme_scratch/edison/mf-edisonhot02/SMS_D_Ln5.ne4_ne4.FC5.edison_intel.20170727_104026_bpn8du

@ndkeen
Copy link
Contributor Author

ndkeen commented Nov 11, 2017 •
edited
Loading

Looking at recent runs from this test, I no longer see that message, but I do see:

00:  Invalid PIO rearranger comm max pend req (comp2io),            0
00:  Resetting PIO rearranger comm max pend req (comp2io) to           64
00:  PIO rearranger options:
00:    comm type     =
00:  p2p                                                                            
00:   
00:    comm fcd      =
00:  2denable                                                                       
00:   
00:    max pend req (comp2io)  =           0
00:    enable_hs (comp2io)     = T
00:    enable_isend (comp2io)  = F
00:    max pend req (io2comp)  =          64
00:    enable_hs (io2comp)    = F
00:    enable_isend (io2comp)  = T

Which may be something different?

example case dir:
/global/cscratch1/sd/ndk/acme_scratch/cori-knl/mf-cons-svml/SMS_D_Ln5.ne4_ne4.FC5AV1C-L.cori-knl_intel.cam-cosplite_nhtfrq5.20171107_165102_70lkp9

@ndkeen ndkeen changed the title PIO warning in acme_developer test RP_Ln9.ne30_ne30.FC5.cori-haswell_intel.cam-outfrq9s PIO warning in acme_developer test RP_Ln9.ne30_ne30.FC5.cori-haswell_intel.cam-outfrq9s Nov 11, 2017
@ndkeen
Copy link
Contributor Author

ndkeen commented Jan 19, 2018

I think we can assume this is not really an issue?

@ndkeen ndkeen closed this as completed Jun 27, 2018
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees

@jayeshkrishna jayeshkrishna

Labels
Cori Machine Files
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
2 participants
@ndkeen @jayeshkrishna