从源码来看ES shard 级别哪些不能命中缓存

最新推荐文章于 2023-10-08 23:30:00 发布

水的精神

最新推荐文章于 2023-10-08 23:30:00 发布

阅读量633

点赞数

CC 4.0 BY-SA版权

分类专栏： elasticsearch源码文章标签： elasticsearch 搜索引擎

本文链接：https://blog.csdn.net/star1210644725/article/details/126608570

elasticsearch源码专栏收录该内容

11 篇文章

订阅专栏

本文探讨了Elasticsearch中滚动查询(scroll)不使用缓存的原因，并分析了影响查询是否使用缓存的因素，包括查询类型、profile分析设置及索引级别的配置等。

摘要生成于 C知道，由 DeepSeek-R1 满血版支持，前往体验 >

本文 es 源码基于 7.11.1

scroll滚动查询的不能缓存
查询类型不是 query than fetch的
开启profile分析的请求
如果该索设置了不缓存 index.requests.cache.enable = false
在请求的时候就设置了不走缓存的

看源码

/**
     * Can the shard request be cached at all?
     */
    public boolean canCache(ShardSearchRequest request, SearchContext context) {
        // Queries that create a scroll context cannot use the cache.
        // They modify the search context during their execution so using the cache
        // may invalidate the scroll for the next query.
        if (request.scroll() != null) {
            return false;
        }

        // We cannot cache with DFS because results depend not only on the content of the index but also
        // on the overridden statistics. So if you ran two queries on the same index with different stats
        // (because an other shard was updated) you would get wrong results because of the scores
        // (think about top_hits aggs or scripts using the score)
        if (SearchType.QUERY_THEN_FETCH != context.searchType()) {
            return false;
        }

        // Profiled queries should not use the cache
        if (request.source() != null && request.source().profile()) {
            return false;
        }

        IndexSettings settings = context.indexShard().indexSettings();
        // if not explicitly set in the request, use the index setting, if not, use the request
        if (request.requestCache() == null) {
            if (settings.getValue(IndicesRequestCache.INDEX_CACHE_REQUEST_ENABLED_SETTING) == false) {
                return false;
            } else if (context.size() != 0) {
                // If no request cache query parameter and shard request cache
                // is enabled in settings don't cache for requests with size > 0
                return false;
            }
        } else if (request.requestCache() == false) {
            return false;
        }
        // We use the cacheKey of the index reader as a part of a key of the IndicesRequestCache.
        assert context.searcher().getIndexReader().getReaderCacheHelper() != null;

        // if now in millis is used (or in the future, a more generic "isDeterministic" flag
        // then we can't cache based on "now" key within the search request, as it is not deterministic
        if (context.getQueryShardContext().isCacheable() == false) {
            return false;
        }
        return true;

    }