Fragments and configs for Herwig7 VBS VW semilep NanoGen

[valsdav]

No description provided.

[kdlong]

The configs look good to me Davide. Were you able to generate samples successfully?

[valsdav]

Hi Kenneth,
Some samples were successful: WpLnuWpJJ, ZToLLZToJJ, WmToLNuWmTo2J.
Others are failing with a seg fault in the middle of the processing https://cmsweb.cern.ch:8443/scheddmon/0155/cms1963/210208_094926:dvalsecc_crab_WminusToJJZTo2LJJ_VBS_Herwig7/job_out.18.0.txt

Do you have any clue maybe about what is happening there?
Thanks!

[kdlong]

No idea. Looks like it's coming from Herwig. What fraction of jobs does it happen in? Does it happen for all the samples you try? Probably worth asking about on the gen hypernews.

[valsdav]

It happens for all the samples apart from WpLnuWpJJ, ZToLLZToJJ, WmToLNuWmTo2J.
But if I run the config locally I don't have problems.. maybe it is a memory limit issue? I'm try increasing the memory on crab..

[kdlong]

Hmm, I guess it's possible, but it seems really weird that it would already have excessive memory use after ~100 events. You could check the memory locally with /usr/bin/time -v cmsRun ... to see if it seems like the memory use is atypical.

[valsdav]

it seems it is taking maximum 758MB of RAM, I don't think it is too much.

It is algo strange that running the config locally it works, running the crab job locally with preparelocal it also crashes..
I will ask on the HN, thanks

[valsdav]

Ok now it works! It was a bug of Herwig not backported on CMSSW_10_6_20. GEN contacts are going to fix this.

[kdlong]

Oops, I overlooked this. Does that mean there is no issue for central production? Can you share working configs with the MC contacts?

[valsdav]

Hi Kenneth,
the problem of the strange crash was about a config for the crab splitting by lumisection. I think it is unrelated to the problems we were having about running Herwig for 2017 and 2016.

PS: I'm going to share on the HN the comparison on NanoGen today