[Question]: Why can the task duration sometimes be seen, while other times it cannot?

[TRYSN666]

Self Checks

• [x] I have searched for existing issues search for existing issues, including closed ones.
• [x] I confirm that I am using English to submit this report (Language Policy).
• [x] Non-english title submitions will be closed directly ( 非英文标题的提交将会被直接关闭 ) (Language Policy).
• [x] Please do not modify this template :) and fill in all the required fields.

Describe your problem

[dosubot[bot]]

Task duration is only shown in the UI when the backend provides valid timing data for a task. This usually happens when a task has fully completed and both start and end times are available. If a task is still running, encounters an error, is canceled, or if the backend doesn't send timing info (for example, due to missing or inconsistent metadata), the UI hides the duration to avoid showing misleading or incomplete information. Conditional rendering in the frontend checks for the presence of this data before displaying it, so you may see duration for some tasks but not others depending on their state and data availability. This behavior is by design to prioritize accuracy and user experience example example.

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}    

[TRYSN666]

Retrieval takes too much time. Is there any way to speed up Retrieval?

This is the time taken for the task ： Query: How to configure DHCP, provide nvue commands###怎么配置dhcp，给出nvue命令

Time elapsed: Total: 851503.6ms Check LLM: 4.1ms Check Langfuse tracer: 0.7ms Bind models: 70.6ms Query refinement(LLM): 1337.6ms Retrieval: 827588.0ms Generate answer: 22502.4ms Token usage: Generated tokens(approximately): 652 Token speed: 28/s @dosu

[dosubot[bot]]

Retrieval in RAGFlow can be slow for several reasons, but there are proven ways to speed it up. The biggest bottlenecks are usually the reranker (especially if running on CPU), RAPTOR enhancement, auto-keyword/auto-question features, and large document databases. Disabling the reranker, RAPTOR, and auto-keyword/auto-question features in your knowledge base config can make a big difference example. If you have a GPU, use it for embedding and reranking—this often cuts retrieval time dramatically example.

For plain text PDFs, switch to the "Naive" parser instead of "DeepDoc" to avoid unnecessary overhead example. Make sure your backend (Elasticsearch, Infinity, etc.) is tuned: increase shards, reduce replicas, and adjust refresh intervals for better performance example. Upgrading hardware (CPU, RAM, disk speed) also helps, especially with large datasets.

Recent RAGFlow versions (with the Infinity engine) have improved retrieval speed, so running the latest release is recommended example. If retrieval is still slow, monitor hardware usage and enable debug logs to pinpoint bottlenecks—low CPU/GPU usage may mean you're waiting on slow I/O or external APIs example.

If your queries are complex or use hybrid retrieval (keyword + vector), lowering the top_k parameter and disabling reranking can help example. You can also use metadata filters to narrow the search space, which was recently added to RAGFlow example.

Let me know if you want help with specific config changes or tuning advice for your backend!

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}    

[TRYSN666]

When ragflow is performing Retrieval, what method does it use? Currently, I haven't enabled the reranker, but Retrieval still takes a long time@dosu

[KevinHuSh]

How many files did you index in total?

[TRYSN666]

How many files did you index in total?

There are currently a total of 20 enabled documents. Strangely, when I only had 14 documents before, the retrieval speed was much faster (about 2-5 minutes for LLM to output results), but when I added 6 documents yesterday, the retrieval speed immediately became very slow (about 15-20 minutes for LLM to output results). Later, I closed the dataset and only kept 2 documents, and the retrieval speed was still very slow @KevinHuSh

[KevinHuSh]

Did you apply some kind of rerank model? They all are parsed with General method, aren't they?

[TRYSN666]

Did you apply some kind of rerank model? They all are parsed with General method, aren't they?

yes，i didnt used any rerank model， and also used the general method for parsing. my files are all in doc and docx formats

[KevinHuSh]

Retrieval: 827588.0ms. This is abnormal. May ES is down?

[TRYSN666]

I started es using "docker-compose up -d es01", then went to test Q&A, and still spent more than 90% of the time during the retrieval process. Meanwhile, after starting es, when I parsed the document with the same parameters, the time still increased several times compared to before

[Magicbook1108]

solved?

[Magicbook1108]

Since there has been no further activity for over three weeks, we will proceed to close this issue. If the problem persists or you have additional questions, please feel free to reopen the issue or create a new one. We’re happy to assist anytime.