Not able to set capabilities in remote environment #8708
-
What happened?We enable RAG and photos uploading. Version Informationroot@chat-ai-dev:/opt/LibreChat# docker images | grep librechat Steps to Reproduce
endpoints:
agents:
capabilities: ["file_search", "tools", "actions"]
What browsers are you seeing the problem on?No response Relevant log outputOnly error appears in remote logs:
'
2025-07-28 07:27:26 error: Failed to load service key from file: /app/api/data/auth.json ENOENT: no such file or directory, open '/app/api/data/auth.json'
'
Errors appears in local logs:
'Error getting all prompt groups MongoServerError: cannot nest $ under $in
at Connection.sendCommand (/app/api/node_modules/mongodb/lib/cmap/connection.js:299:27)
at process.processTicksAndRejections (node:internal/process/task_queues:95:5)
at async Connection.command (/app/api/node_modules/mongodb/lib/cmap/connection.js:327:26)
at async Server.command (/app/api/node_modules/mongodb/lib/sdam/server.js:168:29)
at async AggregateOperation.executeCommand (/app/api/node_modules/mongodb/lib/operations/command.js:76:16)
at async AggregateOperation.execute (/app/api/node_modules/mongodb/lib/operations/aggregate.js:87:16)
at async tryOperation (/app/api/node_modules/mongodb/lib/operations/execute_operation.js:207:20)
at async executeOperation (/app/api/node_modules/mongodb/lib/operations/execute_operation.js:75:16)
at async AggregationCursor._initialize (/app/api/node_modules/mongodb/lib/cursor/aggregation_cursor.js:56:26)
at async AggregationCursor.cursorInit (/app/api/node_modules/mongodb/lib/cursor/abstract_cursor.js:633:27) {
errorLabelSet: Set(0) {},
errorResponse: {
ok: 0,
errmsg: 'cannot nest $ under $in',
code: 2,
codeName: 'BadValue'
},
ok: 0,
code: 2,
codeName: 'BadValue'
}
'ScreenshotsThe same setting in remotely and locally: REMOTELY
LOCALLY
Code of Conduct
|
Beta Was this translation helpful? Give feedback.
Replies: 2 comments 4 replies
-
|
It should be working remotely exactly as it does for you locally, as there would be no inherent difference in deployment environments. Can you explain more about your setup? Are you using the exact same docker images you are building for both environments? Is Redis involved for the remote deployment? It's impossible to tell with the current information supplied. |
Beta Was this translation helpful? Give feedback.
-
|
Thank you very much for your answer!
docker --version
Docker version 27.3.1, build ce12230
CONTAINER REPOSITORY TAG IMAGE ID SIZE
LibreChat-API ghcr.io/danny-avila/librechat-api latest abf6c97ceeb2 422MB
LibreChat-NGINX nginx 1.27.0-alpine 208b70eefac1 18.4MB
chat-meilisearch getmeili/meilisearch v1.12.3 4acfcbbb9e08 95.1MB
chat-mongodb mongo latest a6bda40d00e5 296MB
librechat-rag_api-1 ghcr.io/danny-avila/librechat-rag-api-dev-lite latest 42ef0aa9a8fa 463MB
librechat-vectordb-1 ankane/pgvector latest 956744bd14e9 160MBdocker --version
Docker version 26.1.3, build 26.1.3-0ubuntu1~22.04.1+esm1
Container Repository Tag Image Id Size
---------------------------------------------------------------------------------------------------------
LibreChat-API ghcr.io/danny-avila/librechat-api latest 2c0065bd448a 1.208 GB
chat-meilisearch getmeili/meilisearch v1.12.3 9547aac4aabc 141.4 MB
chat-mongodb mongo latest 3628b824922a 908.5 MB
librechat_rag_api_1 ghcr.io/danny-avila/librechat-rag-api-dev-lite latest 8858d2bd9ff5 1.478 GB
librechat_vectordb_1 ankane/pgvector latest f2c967e41f72 440.4 MB
|
Beta Was this translation helpful? Give feedback.
-
|
Hi everyone, Recently, I've been working on setting up some preset agents. I noticed that the issue I was encountering no longer occurs when I use the interface set from version 0.7.8 in the librechat.yaml file, along with properly configuring the capabilities. Everything seems to be functioning correctly now when tested remotely. 
Thank you all so much for your support and help! My suspicion is that the problem might have originated from an unexpected configuration elsewhere in the librechat.yaml file, which unfortunately doesn’t trigger any warnings in the logs or from the YAML validator. I plan to monitor the setup for a few days, and if it remains stable, I’ll be sure to come back and mark this discussion as resolved. Thanks again! Would you like me to help refine it further or adapt the tone more formally or casually? |
Beta Was this translation helpful? Give feedback.
-
|
Totally feel you on this one. These kinds of remote-vs-local ghosts are the hardest to debug — especially when the YAML looks right but the interface just ghosts the config silently. One of the hidden traps we’ve seen a lot in RAG infra (esp. LibreChat) is what we call the “bootstrap illusion” — the config file loads, the UI renders, but some features are silently skipped due to load timing or capability checks that weren’t fully initialized yet. In your case, it makes perfect sense that switching to We documented this kind of issue in our AI Failure Map under No.15: predeploy-collapse — basically, the agent thinks it’s ready, but infra is still booting or misaligned. Anyway, thanks a ton for writing this all up — your findings saved us an hour of guessing when we ran into a similar symptom with vector tool chains not triggering upload states. If you ever want to explore more of these RAG bootstrap traps, happy to share our internal test harness. Cheers. |
Beta Was this translation helpful? Give feedback.
-
|
Thanks for the invite, I'd love to experience more about it. |
Beta Was this translation helpful? Give feedback.
-
|
You are welcome |
Beta Was this translation helpful? Give feedback.
Hi everyone,
Recently, I've been working on setting up some preset agents. I noticed that the issue I was encountering no longer occurs when I use the interface set from version 0.7.8 in the librechat.yaml file, along with properly configuring the capabilities. Everything seems to be functioning correctly now when tested remotely.
Thank you all so much for your support and help!
My suspicion is that the problem might have originated from an unexpected configuration elsewhere in the librechat.yaml file, which unfortunately doesn’t trigger any warnings in the logs or from the YAML validator.
I plan to monitor the setup for a few days, and if it remains stable, I’ll be sure to come back and…