[Wien] weird error files in scf iterations

[Shu Miao]

Dear users,
I observed weird error files during scf iterations. I ran a complex
parallel calculation with 48 k-vectors. I found the error file for the
running step xx was 4k, instead of empty, saying "ERROR in xx". This was
true for  every step in scf, such as lapw0, lapw1_#, lapw2_# (where # was
the #th k-vector), core, mixer, sumpara. However, after xx was done, this
4k file became empty. After the whole scf was done, all error files were
empty. I also verified it using the TiC example. Is this a normal behavior
of scf? Did anyone observe it before?
Thanks

Shu Miao