I just ran into this crash today. I’m mostly lost as well but I can say that JuMP has nothing todo with it. Also I can reliably reproduce it, although it takes in the order of 20 minutes to get there. Running under gdb didn’t give me a backtrace (“no stack”). I’ll see if I can isolate the last call into Clp and see if that’s sufficient to reproduce the crash on its own.
It was sufficient. Now I can get the crash in a few minutes from
using MathProgBase.SolverInterface, Clp, JLD2, SparseArrays
@load "clpcrash.jld2" data
model = LinearQuadraticModel(ClpSolver())
loadproblem!(model, data...)
optimize!(model)