GP out of memory while testing Sage-6.6

[Andrey Novoseltsev]

Hello,

I don't think I've seen it reported before: running "make ptestlong" on Sage-6.6 after distclean crashed my server last night (Debian Wheezy with 3.16 kernel, 2x Opteron 6272, 16 cores each, 128GB RAM, 64GB swap). Today the machine was unresponsive, on reboot logs have shown killing processes because of out of memory, starting with KVM. Rebooting it and trying to test things again:

...
sage -t --long --warn-long 92.2 src/sage/modular/modform_hecketriangle/hecke_triangle_group_element.py
    [709 tests, 208.50 s]
sage -t --long --warn-long 92.2 src/sage/rings/function_field/function_field.py
**********************************************************************
File "src/sage/rings/function_field/function_field.py", line 59, in sage.rings.function_field.function_field
Warning, slow doctest:
    TestSuite(M).run()  # long time (52s on sage.math, 2012)
Test ran for 737.21 s
    [286 tests, 776.39 s]
sage -t --long --warn-long 92.2 src/sage/schemes/elliptic_curves/lseries_ell.py
^CKilling test src/sage/schemes/elliptic_curves/lseries_ell.py
----------------------------------------------------------------------
sage -t --long --warn-long 92.2 src/sage/interfaces/ecm.py  # 2 doctests failed
Doctests interrupted: 2750/2751 files tested
----------------------------------------------------------------------
Total time for all tests: 844.3 seconds
    cpu time: 17406.4 seconds
    cumulative wall time: 21262.9 seconds
make: *** [ptestlong] Error 129

This was the last untested file (normally tests finish in about 10 minutes). Top process in top before killing:

13972 novoselt  20   0 90.4g  90g 6344 R   100 71.7   4:32.27 /home/novoselt/sage/local/bin/gp --fast --emacs --quiet --stacksize 10000000

Thank you,
Andrey

[Andrey Novoseltsev]

Same story with 6.7.beta3

[John Cremona]

I have seen similar things (also with lseries_ell and gp) and will not
do make testlong or ptestlong until it has been fixed since it is not
fair on other users of the machine. No problem with the build though.

I know that I should help to diagnose and fix this since it is a
functionality which I actually use all the time, but I do not have
time right now.

Jonn

On 29 April 2015 at 22:39, Andrey Novoseltsev <novo...@gmail.com> wrote:
> Same story with 6.7.beta3
>
> --
> You received this message because you are subscribed to the Google Groups
> "sage-devel" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to sage-devel+...@googlegroups.com.
> To post to this group, send email to sage-...@googlegroups.com.
> Visit this group at http://groups.google.com/group/sage-devel.
> For more options, visit https://groups.google.com/d/optout.

[Andrey Novoseltsev]

OK, since this is not just me and it crushes machines:

http://trac.sagemath.org/ticket/18340

[leif]

Andrey Novoseltsev wrote:
> I don't think I've seen it reported before: running "make ptestlong" on

> Sage-6.6 after distclean crashed my server last night.

Similar has been reported to sage-release about two weeks ago:

http://thread.gmane.org/gmane.comp.mathematics.sage.release/2757


-leif

[John Cremona]

That's right, same problem.

John

>
>
> -leif

[leif]

Andrey Novoseltsev wrote:
> OK, since this is not just me and it crushes machines:
>
> http://trac.sagemath.org/ticket/18340

Needs review now (upgrades PARI to a later devel version, fixing the
memory issue).


-leif

[John Cremona]

Excellent! Before I review it, can someone say whether or not it is
relevant what is in a personal $HOME/.gprc file? Mine has

parisizemax = 20000M

John

>
> -leif

[leif]

John Cremona wrote:
> On 2 May 2015 at 18:56, leif <not.r...@online.de> wrote:
>> Andrey Novoseltsev wrote:
>>> OK, since this is not just me and it crushes machines:
>>>
>>> http://trac.sagemath.org/ticket/18340
>>
>> Needs review now (upgrades PARI to a later devel version, fixing the
>> memory issue).
>>
>
> Excellent!

Jeroen's work.

> Before I review it, can someone say whether or not it is
> relevant what is in a personal $HOME/.gprc file? Mine has
>
> parisizemax = 20000M

Shouldn't be relevant; Sage's tests (now) don't use that much... ;-)

(Otherwise you could simply temporarily rename that file.)


-leif

[leif]

P.S.: In the tests where we call 'gp' (rather than use the library), we
use '--fast':

[-f,--fast] Fast start: do not read .gprc


-leif