[LON-CAPA-admin] Serious troubles with lonmaxima

[Stefan Bisitz]

Guy Albertelli wrote:
> Hi Stefan,
> 
> Hrrm, I haven't been to recreate this yet here, perhaps there might be
> an example HW that you have that seems to cause or be more likely
> related to either of these failure conditions? 
> 
> I've done some minor testing here. (Basically just getting one
> homework problem that does some maxima interactions to see if I could
> cause either condition to occur.)
> 
> I've done several thousands of requests and seen the maxima
> children go through multiple clean restarts on both fc5 and SLES9 and
> haven't yet seen either failure condition occur.
> 
> Not sure what else I can try doing here...

Hi Guy,

We have done some more testing and are willing to do even more to check
some things. We can grant access to one of our servers for you lateron.
do you need full access (Webinterface) or is ssh connection enough?

The maxima logfile looks fine. Maybe, we add some more things to be logged.

I updated our development server (SLES9 32 Bit, LON-CAPA 2.4.2). It
shows similar reactions: At least I got it twice to do the 10 second
break just by clicking "New Randomization" several times in a problem
which uses &cas. The corresponding lisp processes (99% CPU for some
seconds) have stopped automatically.

Anyway, we have to be carful with old maxima versions which I installed
by hand months ago for testing. The LON-CAPA update does not totally
remove or deactivate these old versions. But on the new server (FC5 64
Bit, 2.4.1) which shows the same problems, there is only Maxima 5.12.0.
Strange...

Let us do some more tests.

Stefan Bisitz