Django rest framework 简单模型序列化器列表视图永远显示 10 条记录

Django rest framework simple model serializer list view takes forever to show 10 records

我有这个模型 -

class News(BaseEntityBasicAbstract, HitCountMixin):
    """
    News added from the dashboard with content
    """
    NEWS_STATUS = (
        ('draft', _('Draft')),
        ('pending', _('Pending')),
        ('review', _('Review')),
        ('public', _('Public')),
        ('private', _('Private'))
    )
    backup = models.BooleanField(default=False)
    prev_id = models.BigIntegerField(null=True, blank=True)
    language = models.CharField(max_length=10, choices=LANGUAGES, default='bn', db_index=True)
    heading = models.CharField(max_length=255, null=True, blank=True,
                               verbose_name=_('News Heading'),
                               help_text=_('Provide a news heading/caption.'))
    sub_caption = models.TextField(max_length=255, null=True, blank=True,
                                   verbose_name=_('Summary'),
                                   help_text=_('Provide summary of the news.'))
    url = models.CharField(max_length=255, unique=True, verbose_name=_('URL/Slug/Link'),
                           help_text=_('Unique url for the news without whitspace.'))
    content = HTMLField(null=True, blank=True, verbose_name=_('Content'),
                        help_text=_('HTML content with texts, links & images.'))
    featured_image = models.FileField(upload_to=FilePrefix('news/'), null=True, blank=True,
                                      verbose_name=_('Featured Image'),
                                      help_text=_('Upload a featured image for news.'))
    image_caption = models.TextField(max_length=255, null=True, blank=True,
                                     verbose_name=_('Image Caption'),
                                     help_text=_('Provide a image caption.'))
    status = models.CharField(max_length=20, choices=NEWS_STATUS, default='pending',
                              verbose_name=_('News Status'), db_index=True,
                              help_text=_('Only public news can be seen on front end.'))
    source = models.ForeignKey(NewsSource, on_delete=models.SET_NULL, null=True, blank=True,
                               verbose_name=_('News Source'),
                               help_text=_('Select a news source.'))
    category = models.ForeignKey(Category, on_delete=models.SET_NULL, null=True, blank=True,
                                 verbose_name=_('Category'),
                                 help_text=_('Select a news category.'))
    tags = tagulous.models.TagField(
        blank=True,
        to=Tags,
        verbose_name=_('News Tags'),
        help_text=_('Provide news tags separated with commas.')
    )
    published_at = models.DateTimeField(null=True, blank=True, db_index=True,
                                        verbose_name=_('Published At'))
    menu_items = GenericRelation(MenuItems, object_id_field='id',
                                 related_query_name='news_as_menu')
    hit_count_generic = GenericRelation(HitCount, object_id_field='object_pk',
                                        related_query_name='news_hit_count')
    created_by = models.ForeignKey(User, related_name='news_created_by',
                                   on_delete=models.SET_NULL, null=True, blank=True,
                                   verbose_name=_('Created By'))
    updated_by = models.ForeignKey(User, related_name='news_updated_by',
                                   on_delete=models.SET_NULL, null=True, blank=True,
                                   verbose_name=_('Last Updated By'))
    published_by = models.ForeignKey(User, related_name='news_published_by',
                                     on_delete=models.SET_NULL, null=True, blank=True,
                                     verbose_name=_('Published By'))
    deleted_by = models.ForeignKey(User, related_name='news_deleted_by',
                                   on_delete=models.SET_NULL, null=True, blank=True,
                                   verbose_name=_('Deleted By'))

下面是序列化程序 -

class NewsSerializer(serializers.ModelSerializer):
    class Meta:
        model = News
        fields = ['id', 'heading', 'sub_caption', 'url', 'content', 'featured_image',
                  'image_caption', 'category', 'source', 'tags', 'published_at']

这是视图 -

class NewsViewSets(viewsets.ModelViewSet):
    queryset = News.objects.filter(
        is_active=True,
        status='public'
    )
    serializer_class = NewsSerializer

    def get_queryset(self):
        queryset = self.queryset.filter(
            language=self.request.LANGUAGE_CODE
        ).order_by('-id')

        return queryset

分页在settings.py中设置为仅10,当我点击新闻api url时,仅加载10条记录需要9/10秒。这是显示 django-debug-toolbar 报告的屏幕截图 -

我在数据库中有大约 40 万条记录 table,这可能是个问题,但我认为加载时间太长了。请帮我找到这里的问题!提前致谢。

过滤通常很慢。看起来您在相关字段上有数据库索引,但请注意,如果您有多个过滤器,则只会使用其中一个索引。

我是根据您的专栏猜测的,但似乎最常见的查询总是寻找 is_active=1status='public'。如果不是这种情况,您可能需要进行一些调整。

首先,删除 statusis_activelanguage 字段上的 db_index=True,否则您的数据库写入会不必要地变慢。

那么你可以制定这样的索引:

class Meta:
    indexes = [
        models.Index(
            fields=["is_active", "status", "language"],
            name="idx_filtering",
        )
    ]

当您一次过滤所有三列时,这将有助于数据库。但是,如果您只对其中一列进行过滤,您可能希望保留原始 db_index=True.

如果您使用的是 PostgreSQL,您可以做得更好:

class Meta:
    indexes = [
        models.Index(
            fields=["is_active", "status", "language"],
            name="idx_filtering",
            condition=Q(is_active=True, status="public"),
        )

这会将索引的大小减小到仅匹配 Q() 的索引,从而加快遍历速度。

另一件需要注意的事情是,一旦你达到更高的偏移量,使用 OFFSET 的分页就会非常慢。如果可能的话,您应该改用 DRF 的光标分页。