FC5AV1C-04P2.ne30_ne30 failing on cori-knl and cori-haswell

[xuezhengllnl]

The error message in e3sm.log:
2680: forrtl: severe (174): SIGSEGV, segmentation fault occurred
2680: Image PC Routine Line Source
2680: e3sm.exe 0000000004487CED Unknown Unknown Unknown
2680: e3sm.exe 0000000003CF7600 Unknown Unknown Unknown
2680: e3sm.exe 0000000001916FA0 radar_simulator_ 230 radar_simulator.f90
2680: e3sm.exe 000000000190DB04 mod_cosp_radar_mp 172 cosp_radar.F90
2680: e3sm.exe 00000000017D3805 mod_cosp_simulato 105 cosp_simulator.F90
2680: e3sm.exe 0000000001496072 mod_cosp_mp_cosp_ 615 cosp.F90
2680: e3sm.exe 0000000001492BEB mod_cosp_mp_cosp_ 194 cosp.F90
2680: e3sm.exe 0000000001049AB4 cospsimulator_int 3922 cospsimulator_intr.F90
2680: e3sm.exe 00000000007B364A radiation_mp_radi 1572 radiation.F90
2680: e3sm.exe 0000000001187706 physpkg_mp_tphysb 2747 physpkg.F90
2680: e3sm.exe 0000000001181111 physpkg_mp_phys_r 1047 physpkg.F90
2680: e3sm.exe 00000000005006B7 cam_comp_mp_cam_r 250 cam_comp.F90
2680: e3sm.exe 00000000004EC929 atm_comp_mct_mp_a 577 atm_comp_mct.F90
2680: e3sm.exe 000000000042D0DF component_mod_mp_ 737 component_mod.F90
2680: e3sm.exe 000000000040CFF7 cime_comp_mod_mp_ 2790 cime_comp_mod.F90
2680: e3sm.exe 000000000042CD5D MAIN__ 133 cime_driver.F90
2680: e3sm.exe 000000000040A80E Unknown Unknown Unknown
2680: e3sm.exe 000000000459AA79 Unknown Unknown Unknown
2680: e3sm.exe 000000000040A6F9 Unknown Unknown Unknown

[ndkeen]

I counter with these 2 passing tests on cori-knl using master as of today (july 22):

create_test SMS_Ln5.ne30_ne30.FC5AV1C-04P2 SMS_D_Ln5.ne30_ne30.FC5AV1C-04P2

[lxu16]

I met the same error last week on cori. Uou might want to try switch off "-cosp" in your runscript to see if it succeed?

[xuezhengllnl]

Thanks @lxu16
I tried to run it without "-cosp" and it worked, which is consistent with the error message. However, I need to switch on "-cosp" to get the cosp simulator output for the model diagnostics.

[lxu16]

@xuezhengllnl Did you try to define cosp_lite=.true. in the namelist and use "-cosp" in the runscript? Just to check it works and the output variables you need for the model diagnostics.

[xuezhengllnl]

@xuezhengllnl Did you try to define cosp_lite=.true. in the namelist and use "-cosp" in the runscript? Just to check it works and the output variables you need for the model diagnostics.

@lxu16 It is working with cosp_lite. Thanks!