The mailing list for listmasters using Sympa

[Olivier Salaün <address@concealed>]

Hi Stefan,

Thank you for your quick feedback :-)

Le 04/05/2020 à 12:27, Stefan Hornburg (Racke) a écrit :
> On 5/4/20 11:33 AM, Olivier Salaün wrote:
> > Hello,
> >
> > I wish to summarize here the results of our investigation to understand the 
> > origin of very bad performances of wwsympa
> > in our production server (Sympa 6.2.48).
> >
> > [...]
> Hello Olivier,
>
> thanks a lot for your report. This is certainly a sore point in the usability 
> of the web interface, so improvements
> there would be much appreciated.
>
> Comments:
>
> #1 This patch looks good to me, so it would be good to run tests on it and 
> merge it when we are satisfied with
>     the result.

I did some tests to ensure this patch does not prevent param_constraint 
checking while instantiating, modifying, adding lists to families. My 
tests (based on Test::BDD::Cucumber) are public on 
https://github.com/salaun-urennes1/sympa-functional-tests

> #5 I think locking is not necessary for compiling the list information.

It probably makes sense to still do a shared lock while reading 
configuration files to prevent another process that need to write to get 
an exclusive lock.

But File::NFSLock is probably the culprit: using another more simple and 
efficient file locking, when NFS is not used, could make us save some 
processing time.

--
Olivier Salaün
DSI / pôle SI / équipe SNUM
Tel : 02 23 23 74 54