What causes "Fine-tuning model does not exist" in iflytek/astron-agent?

Verify the modelId actually refers to a fine-tuning model

iflytek/astron-agent · error · BusinessException

RESPONSE_FAILED

RESPONSE_FAILED

Error message

Fine-tuning model does not exist

What it means

LLMService's fine-tuning enable/update path keeps a cached map (catchModelMap, also persisted in Redis) of known fine-tuning model ids. If the requested modelId is not present in the cache, it throws BusinessException(RESPONSE_FAILED, "Fine-tuning model does not exist"): the model was never registered as a fine-tuning model (or the cache was lost/reset).

Solutions

Verify the modelId actually refers to a fine-tuning model
Check the Redis key holding the cached map; if missing, re-trigger the cache population (model list sync) before toggling
Re-sync the fine-tuning model cache from the upstream source and retry
Fix the misleading generic RESPONSE_FAILED enum by using a dedicated not-found enum

Example fix

// before
Boolean exists = catchModelMap.get(key);
if (exists == null) {
    throw new BusinessException(ResponseEnum.RESPONSE_FAILED, "Fine-tuning model does not exist");
}
// after
Boolean exists = catchModelMap.get(key);
if (exists == null) {
    refreshFineTuningModelCache(); // re-populate from source before failing
    exists = catchModelMap.get(key);
}
if (exists == null) {
    throw new BusinessException(ResponseEnum.MODEL_NOT_EXIST, "Fine-tuning model does not exist: " + key);
}

Defensive patterns

Strategy: fallback

Validate before calling

// check cache membership before toggling
Boolean known = redisTemplate.opsForValue().get(fineTuningCacheKey) != null;
if (!known) refreshFineTuningModelCache();

Try / catch

try {
    llmService.setFineTuningModelEnabled(modelId, enable);
} catch (BusinessException e) {
    if (e.getMessage() != null && e.getMessage().contains("Fine-tuning model does not exist")) {
        refreshFineTuningModelCache(); // then retry once
    }
}

Prevention

Persist and rehydrate the fine-tuning model cache on startup
Avoid Redis flushes in shared environments or re-populate after
Only pass model ids sourced from the fine-tuning model list
Use a dedicated not-found enum instead of RESPONSE_FAILED + message

When it happens

Trigger: Toggling/enabling a fine-tuning model by id when catchModelMap has no entry for that id — e.g. after a Redis flush, service restart with empty local cache, or passing a regular (non fine-tuning) model id.

Common situations: Redis flushed or evicted the cached map key; environment switch (cache populated in dev, request in prod); caller passing a non-fine-tuning model id; cache population step failed earlier and was ignored.

Understand the failure class

Background: Record Not Found Errors: "not found", RecordNotFound, and "was not found" — what they mean and how to fix them — this error's family across 28 libraries.

Related errors

AI-assisted analysis of iflytek/astron-agent@5e758547a8 (2026-09-12). Data as JSON: /api/errors/f17f6646588bcea1. Report an issue: GitHub.

Appendix: source

Thrown at console/backend/toolkit/src/main/java/com/iflytek/astron/console/toolkit/service/model/LLMService.java:588

        return ApiResult.success(configs);
    }

    /**
     * Whether to enable fine-tuning model
     *
     * @param modelId
     * @param enable
     */
    public void switchFinetuneModel(Long modelId, Boolean enable) {
        final String enabledKey = MODEL_ENABLE_KEY.concat(UserInfoManagerHandler.getUserId());
        Map<String, Boolean> catchModelMap = getCatchModelMap(enabledKey);
        if (catchModelMap == null) {
            catchModelMap = new HashMap<>();
        }
        final String key = String.valueOf(modelId);
        Boolean exists = catchModelMap.get(key);
        if (exists == null) {
            throw new BusinessException(ResponseEnum.RESPONSE_FAILED, "Fine-tuning model does not exist");
        } else {
            catchModelMap.put(modelId.toString(), enable);
            redisTemplate.opsForValue().set(enabledKey, JSON.toJSONString(catchModelMap));
        }
    }
}

View on GitHub (pinned to 5e758547a8)