On May 8, 12:03 am, Daniel Roseman <roseman.dan...@googlemail.com>
wrote:
> On May 7, 5:56 pm, Eric Chamberlain <e...@rf.com> wrote:
>
> > Our view is taking too long to render, can anyone offer suggestions to
> > improve its speed?
>
> > There are 5830 profiles.
>
> > Total query count: 23809
> > Total duplicate query count: 493
> > Total SQL execution time: 66.003
> > Total Request execution time: 142.931
>
> <snip models>
>
>
>
>
>
>
>
> > profiles =
> > Profile
> > .objects
> > .filter(partner=request.partner,user__is_active=True).order_by('-
> > user__date_joined')
> > for p in profiles:
> > p.provider_list = list(
> > account.provider for account in
> > ProviderAccount
> > .objects
> > .filter(user=p.user,provider__provider_partners=request.partner))
> > p.call_count = p.call_records.filter().count()
> > return PartnerResponse(request, {'profiles': profiles})
>
> > PartnerResponse returns the view to the user, the template uses a for
> > loop to iterate through the profiles.
>
> As it happens, reducing query counts is something of an obsession of
> mine at the moment - just gave a talk at EuroDjangoCon on this very
> subject.
>
> In your case, the first thing to do is take that ProviderAccount query
> out of the loop. What I would do is first get a list of all provider
> user_ids, then query all ProviderAccounts with those users. You can
> then put them into a dictionary keyed by user_id and then loop through
> to re-associate them with the profile.
>
> Something like this (untested):
>
> profiles = Profile.objects.filter(partner=request.partner,
> user__is_active=True).order_by('-user__date_joined')
> user_ids = [p.user_id for p in profiles]
>
> accounts = ProviderAccount.objects.filter(user__in=user_ids,
> provider__provider_partners=request.partner)
> account_dict = dict([(pa.user_id, pa) for pa in accounts])
>
> for p in profiles:
> p.provider_list = account_dict[p.user_id]
>
> You will probably need to tweak this - it doesn't take account of the
> fact that a user might be associated with multiple ProviderAccounts -
> but hopefully you get the idea, you should try and get all your
> accounts in one big query and *then* associate them with the relevant
> profiles.
>
> The other thing to look at is the proper use of select_related. You
> might find that doing Profile.objects.select_related().filter
> (whatever) has a big impact on reducing the count.
>
> Finally, you have not stated what call_records is, it doesn't appear
> to come from anywhere in this code, so I don't know if that's also the
> cause of some big queries.
>
> --
> DR.
And, just to follow up on my post, the way to handle multiple accounts
for each provider is to do this instead of the list comprehension I
posted above:
account_dict = {}
for pa in accounts:
account_dict.setdefault(pa.user_id, []).append(pa)
--
DR.
--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups
"Django users" group.
To post to this group, send email to django-users@googlegroups.com
To unsubscribe from this group, send email to
django-users+unsubscr...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/django-users?hl=en
-~----------~----~----~----~------~----~------~--~---
