Skip to content

Monitoring: Proactive LLM_CALL checks #897

@Prajna1999

Description

@Prajna1999

Is your feature request related to a problem?
We currently lack proactive monitoring for the llm_call surface area (model, provider combinations). This leaves us unaware of issues until clients report them.

Describe the solution you'd like

  • Implement a CRON JOB to perform dummy calls to LLM_CALL with varying payloads across providers, models, and modalities in a round-robin fashion.
  • Check the database for failures.
  • Log any failures as Sentry alerts on Discord.
Original issue

Describe the current behavior
We are in blind if llm_call surface area (model, provider combinations) are not healthy. Only after clients complaint we start investigating.

Describe the enhancement you'd like
CRON JOB --> LLM_CALL with dummy payload (across providers, models, modalities etc) in round robin--> Check DB --> if something fails --> shows up in Sentry alerts on Discord.

Why is this enhancement needed?
Proactively keep a tab on llm/call surface.

Metadata

Metadata

Assignees

Labels

enhancementNew feature or request

Type

No type
No fields configured for issues without a type.

Projects

Status
To Do

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions