Computation termination because of OOM is not traced as such

It has been observed that computation can terminate with a log:

Terminate computation: <<computation name>> due to previous failure

but looking at the previous server.log trace there is no error related to the computation thread.
On the other hand, the server.log contains OutOfMemory errors on other threads at the same time,
it sounds that such OOM is not caught/traced by the ComputationRunner thread or failsafe mechanism.