I updated to julia 1.9.3, and I am unable to run my code on a cluster.
The 1st issue is the precompilation. It seems to me that every time I open a new session with the cluster, I need to precompile my project. As far as I understand, this is a julia 1.9 thing. Is there a way to deal with this when working on a cluster?
The 2nd issue, is a segmentation fault.
Running my script I see these warnings:
<E2><94><8C> Warning: The call to compilecache failed to create a usable precompiled cache file for FileIO [5789e2e9-d7fb-5bc7-8068-2c6fae9b9549]
<E2><94><82> exception = ArgumentError: Invalid checksum in cache file /home/labs/orenraz/roiho/.julia/compiled/v1.9/FileIO/6iKRU_gncnE.so.
<E2><94><94> @ Base loading.jl:1818
and
<E2><94><94> @ Base loading.jl:1793
<E2><94><8C> Warning: Module FileIO with build ID ffffffff-ffff-ffff-001c-a130a3b6bf6f is missing from the cache.
<E2><94><82> This may mean FileIO [5789e2e9-d7fb-5bc7-8068-2c6fae9b9549] does not support precompilation but is imported by a module that does.
Followed by
/scratch/1697361899.830137.65: line 8: 58683 Segmentation fault (core dumped)
I posted this question before, but I phrased it badly, and that is why I opened this new thread.
I’d love to know if there is a “good” way to run julia 1.9 on a cluster or if the best thing is to go back to julia 1.7 (which is the version I had before and was working well).