Been trying some semi-automagical-load-balancing. Manodeep suggested I focus on this PR. I was able to do a few runs successfully but the ones today have failed when ones from a few days ago worked.
Here's the (AI assisted) theory..
Test01/02/03 ran while the old ESMF spack hash still existed on disk. The binary (access-om3-MOM6-CICE6 from access-om3/pr236-2) was compiled against ESMF at:
/g/data/vk83/apps/spack/1.1/release/linux-x86_64_v4/esmf-8.9.1-u56mitgtfbi2qlj3x7lw7ofxj7lw4b2l/lib/libesmf.so
This got removed overnight as part of this #model-release > Retracting A Release @ 💬
@manodeep got me to check the shared libraries for the pre-release build:
[cyb561.gadi-login-05: dev-MC_25km_jra_iaf+wombatlite_test06_26nodes_216nonOCN]$ ldd which access-om3-MOM6-CICE6 linux-vdso.so.1 (0x00007fffff9db000) /g/data/vk83/apps/spack/1.1/release/linux-x86_64_v4/esmf-8.9.1-u56mitgtfbi2qlj3x7lw7ofxj7lw4b2l/lib/libesmf.so => not found libpiof.so => not found libpioc.so => not found
So I think that's consistent with Claude suggestion but I don't have a timemachine to go back run this command when it was working. So @CodeGat has suggested re-deploying the Pre-release build.
Described:
Been trying some semi-automagical-load-balancing. Manodeep suggested I focus on this PR. I was able to do a few runs successfully but the ones today have failed when ones from a few days ago worked.
Here's the (AI assisted) theory..
This got removed overnight as part of this #model-release > Retracting A Release @ 💬
@manodeep got me to check the shared libraries for the pre-release build:
So I think that's consistent with Claude suggestion but I don't have a timemachine to go back run this command when it was working. So @CodeGat has suggested re-deploying the Pre-release build.
Described: