Hi

Slávek Banko píše v Pá 23. 11. 2018 v 15:42 +0100:
> Dne pá 23. listopadu 2018 Michal Čihař napsal(a):
> > > By the way, I had to set "--concurrency=1", because the
> > > individual
> > > worker
> > > will quickly load 790 MiB RAM after the startup and the memory
> > > consumption then rises steadily. Now I was restarting the celery
> > > when
> > > the
> > > consumption was over 1600 MiB RAM. Is this normal behavior?
> > 
> > The workers on Hosted Weblate service consume around 150 MiB RAM,
> > so
> > that is probably not expected. How big is your fulltext index?
> > 
> > Anyway there seem to be some memory leaks in Whoosh, at least there
> > is
> > known issue with deleted documents:
> > 
> > https://bitbucket.org/mchaput/whoosh/issues/492/unnecessary-and-leaked-
> > memory-usage-of
> 
> $ du -chs data/whoosh/
> 3,2M    data/whoosh/
> 
> It seems to me too small. Definitely less than I expected.
> 
> From description of bug I do not know if I should try to rebuild all
> the 
> indexes or what else.

It's just the index, it doesn't contain the strings, but this indeed
looks quite small. You can use following command to rebuild it from
scratch:

./manage.py rebuild_index --all --clean

-- 
        Michal Čihař | https://cihar.com/ | https://weblate.org/

Attachment: signature.asc
Description: This is a digitally signed message part

_______________________________________________
Weblate mailing list
[email protected]
https://lists.cihar.com/cgi-bin/mailman/listinfo/weblate

Reply via email to