Re: question on trouble shooting gp pari sessions which suddenly quit ru

Hello,

This happened to me when I was working with very (too) large vectors. At some point, gp would stop without an error message. I eventually identified (how, I don't remember) that it was a memory problem.

Emmanuel Royer
Professeur à l'Université Clermont Auvergne
https://royer.perso.math.cnrs.fr
----
Institut CNRS-Pauli
IRL2842, Wien, Österreich
CNRS & Wolfgang Pauli Institut

De: American <website.reader3@gmail.com>
à: pari-users <pari-users@pari.math.u-bordeaux.fr>
Envoyé: mardi 23 avril 2024 19:57 CEST
Sujet : question on trouble shooting gp pari sessions which suddenly quit running

Hello:

It is a bit difficult to explain to you all, because I could actually
not find enough information to point me in the right direction to
troubleshoot a problem I've not seen before.

Yesterday afternoon I kicked off a script of 6 programs running a
gp-pari program for each of the 6 cores I have on my workstation for
maximum effort. Each script was processing (or was supposed to process)
around 137K points. Oddly enough while I was sleeping last night, all 6
programs quit running, and when I checked this morning using the "ps"
command, no gp pari program was running at all. All 6 had terminated
after each one processing about 22K points and that was that.

How could a person go about troubleshooting this? Right now it appears
that the programs died on their own. The GP Pari command scripts were
also checked using the gp2c-run command and they appeared to pass the C
warnings flags, with no errors at all, so I am puzzled as to what really
happened. I checked the warn log in the /var/logs area, and nothing
shows up pertaining to gp-pari from yesterday and today. A few days ago,
the warn log showed a print malloc error and terminated the run but this
might have been my manual work.

I do have print statements embedded in the main program, but never found
them to be a problem like this before.

I am curious as to how to find out what is really happening here? I
restarted all 6 programs again, and I suppose when the count of output
reaches around 22K again, I could attempt to carefully monitor them.

Randall