Knowledge base retrieval aborted. Request took too...
# support
s
I deleted knowledge created again but result is some. Why ? Please Help! [LOG] Knowledge base retrieval aborted. Request took too long. Waited for 750 ms
d
Hi Samet, I can help out with the issue
Check your dm
@Samet Şimşek
s
hi
a
Hey! To help track down this issue, could you share: - The call ID - When exactly this happened (the timestamp) - What response you expected to get - What response you actually got instead This would really help us figure out what went wrong!
s
Time: May 20, 2025, 10:31 callid: d0188511-44e2-48ad-b937-c604b76a3109 I am asking it questions that could be answered from the knowledge stored in Trieve.ai. However, due to this error, it can’t respond and no answer is returned. I keep seeing the same error in the logs. 07:31:25:187 [LOG] Knowledge base retrieval aborted. Request took too long. Waited for 750 ms
a
It happened because Trieve took a long time to respond, so we had to abort it.
s
I think this issue is not caused by Trieve but by the prompt itself. Its length was probably causing the problem — I haven’t seen this error when using a shorter prompt. It’s working now. I believe adding a system prompt length limit would be a good idea, as I think that’s where the issue originated.
a
Got it, that's an interesting point. I'll keep an eye on it.
s
Can we increase timout for trieve connection ? Is there any params on vapi asistant ?
a
No, we can't increase it.
s
@Nick K - CEO @ Trieve might be able to suggest you something why it is taking so much time to respond back.
n
Likely was due to prompt length as Samet found 👍
s
The problem is not the prompt length, guys, I started experiencing the same thing with the short prompt. I tried adding the instructions one by one, changed them and tried again. I finally agreed that the real problem was the late response from Trieve. When I went into detail, the problem was solved when I changed the embedding model in the data set settings. Trieve gives results in a shorter time. But still, as the data grows, it would be great for Vapi to set a delay parameter so that similar problems do not occur tomorrow. Thank you for your interest.