IOError : [Errno 2] No such file or directory: '.../GX.X/results.dat'

Dear developers,

I'm performing subsequent runs (on CP3's cluster) for the same processes (decays) with different parameters and, once in a while (a few/10 runs), I get something like
---
Idle: 0 Running: 6 Finish: 106
INFO: All jobs finished
Command "calculate_decay_widths squarkLSP_200_750_2 --cluster" interrupted with error:
IOError : [Errno 2] No such file or directory: '/nfs/home/fynu/gdurieux/SSDL_8t
ev/GEN_RPV_SIMP_widths_squarkLSP/SubProcesses/P0_ul_udxsxtx/G5.8/results.dat'
---
I cannot check if the file is really present because, when I look at it, I think it is replaced by the one of the next run.

The debug.log file indicates this occurs in filling the html files:
---
  File "madgraph/interface/madevent_interface.py", line 2505, in do_survey
    cross, error = sum_html.make_all_html_results(self)
  File "madgraph/various/sum_html.py", line 384, in make_all_html_results
    P_comb.add_results(name, pjoin(P_path,name,'results.dat'), mfactor)
  File "madgraph5/madgraph/various/sum_html.py", line 124, in add_results
    oneresult.read_results(filepath)
  File "madgraph5/madgraph/various/misc.py", line 155, in deco_f_retry
    return f(*args, **opt)
  File "madgraph5/madgraph/various/sum_html.py", line 56, in read_results
    for line in open(filepath):
---

In GX.X folder, the <my-run>_log.txt and <my-run>_results.dat are absent.
In Events folder, no <my-run> directory is found.

I guess this is not really a MG issue (other runs seem fine) but since this occurs quite frequently, I was wondering if there could be a workaround.
I've sometimes seen (for the fist run on which the error occurs?), the additional lines:
---
Start waiting for update on filesystem. (more info in debug mode)
fail to do <function read_results at 0x14a8ce60> function with <madgraph.various.sum_html.OneResult object at 0x14d02990>, GX.X/results.dat args. 1 try on a max of 5 (20 waiting time)
---
in the error message. Should I just try to increase the hard-coded "nb_try" parameter of "multiple_try" near line 51 in sum_html.py? Or is there something more subtle that could be tried?

Thanks a lot,
Gauthier