ITADN

Recurring 502 PluginInvokeError on both Anthropic and Gemini LLM nodes (Dify Cloud, since Wed Jul 15)

#39194Opencoo-07 创建于 2026-07-17
bugproject#dify
C
coo-07commented
### Self Checks - [x] I have read the [Contributing Guide](https://github.com/langgenius/dify/blob/main/CONTRIBUTING.md) and [Language Policy](https://github.com/langgenius/dify/issues/1542). - [x] This is only for bug report, if you would like to ask a question, please head to [Discussions](https://github.com/langgenius/dify/discussions/categories/general). - [x] I have searched for existing issues [search for existing issues](https://github.com/langgenius/dify/issues), including closed ones. - [x] I confirm that I am using English to submit this report, otherwise it will be closed. - [x] 【中文用户 & Non English User】请使用英语提交,否则会被关闭 :) - [x] Please do not modify this template :) and fill in all the required fields. ### Dify version Cloud(latest) ### Cloud or Self Hosted Cloud ### Steps to reproduce 1. Chatflow with an IF/ELSE node checking `files` is not empty 2. On the IF branch: a Knowledge Retrieval node, followed by two parallel LLM nodes ("LLM" = Claude Sonnet 5, "LLM2" = Gemini 3.1 Pro Preview), each with retry-on-failure (3 attempts, 1000ms interval) and a fail branch configured 3. Send a chat message with 3 image attachments (PNG, ~2.5-3MB each) plus a text prompt (~2000 characters) 4. Both LLM nodes intermittently fail with the above 502 error after exhausting all 3 retries 5. This does not happen on every run - it's intermittent, but has occurred multiple times per day since July 15 ### ✔️ Expected Behavior The LLM nodes (Claude Sonnet 5 and Gemini 3.1 Pro Preview) should successfully process the request and return generated text. ### ❌ Actual Behavior Both LLM nodes intermittently fail with the following error after exhausting all 3 retries: PluginInvokeError: {"args":null,"error_type":"PluginDaemonInnerError","message":"Error sending request to serverless: all 3 attempts failed, last error: attempt 3/3 failed with status code: 502"} req_ids observed (all times JST, UTC+9): - a20a4aff6b - cc12da581c - a2241d9687 - d6971f5ea8 (2026-07-17, ~10:13 JST) - dc4a6424b4 (2026-07-17, ~14:00 JST) On at least one occasion, both the Anthropic node and the Gemini node failed with the exact same req_id at the same time, which suggests the issue may be on the plugin-daemon / serverless invocation layer rather than with either model provider individually. Public status page (https://dify.statuspage.io) showed all systems operational during these failures.
1 条评论