ErrLookup › vllm-project/vllm
vllm-project/vllm
A high-throughput and memory-efficient inference and serving engine for LLMs · Python · 2,614 source files
Analyzed at c794754062 on 2026-08-14. 553 documented errors.
| Code / Message | Type | Severity | Tags |
|---|---|---|---|
| HTTP request failed: {0} | exception | error | moriio, kv-transfer, validation, numpy, vllm |
| JSON error: {0} | exception | critical | moriio, rdma, kv-transfer, memory-registration, vllm |
| Tokenizer error: {0} | exception | critical | moriio, rdma, timeout, kv-transfer, vllm |
| tokenize endpoint unavailable: {0} | exception | error | moriio, zmq, protocol-version, kv-transfer, concurrency, vllm |
| Configuration error: {0} | exception | error | nixl, eplb, initialization, vllm |
| Endpoint not ready after {0}s: {1} | exception | error | nixl, eplb, elastic-ep, metadata-mismatch, vllm |
| Backend error: {0} | exception | error | nixl, eplb, elastic-ep, dtype-mismatch, vllm |
| IO error: {0} | exception | critical | nixl, eplb, rdma, transfer-failure, vllm |
| Image generation should not fail | panic | error | eplb, pynccl, nccl, device-mismatch, vllm |
| chat request must contain at least one message | exception | error | eplb, pynccl, dtype, quantization, vllm |
| cannot continue the final message when the last message is n | exception | error | eplb, pynccl, nccl, vllm |
| chat template is required but none was configured | exception | error | eplb, pynccl, initialization, vllm |
| chat template error: {0} | exception | error | eplb, elastic-ep, stateless, backend-selection, vllm |
| multimodal input is not supported by this chat renderer | validation | error | nixl, eplb, dependency, backend-selection, vllm |
| unsupported multimodal content: {0} | validation | error | nixl, eplb, device-mismatch, vllm |
| `{modality}` input is not supported by this model | validation | error | nixl, eplb, initialization, vllm |
| At most {limit} {modality}(s) may be provided in one prompt. | validation | error | eplb, backend-selection, invalid-argument, vllm |
| multimodal preprocessing error: {0} | exception | error | eplb, model-config, moe, vllm |
| {kind} parsing is not available for model `{model_id}` | exception | error | kv-events, zmq, endpoint, config, vllm |
| {kind} parsing is disabled by frontend configuration | exception | error | kv-events, publisher, registry, config, vllm |
| {kind} parser `{name}` is not registered{} | exception | error | rust, parser, configuration, registry, structured-output |
| failed to initialize {kind} parser `{name}` | exception | error | rust, parser, initialization, wrapped-error |
| gpt_oss uses native Harmony output parsing; generic {kind} p | exception | error | rust, gpt-oss, harmony, parser, configuration |
| harmony output parsing failed | exception | error | rust, gpt-oss, harmony, parsing, streaming |
| this model's maximum context length is {max_model_len} token | validation | error | rust, context-length, validation, tokenizer, chat |
| chat request stream `{request_id}` closed before terminal ou | exception | error | rust, streaming, sse, lifecycle, chat |
| tool call stream state is inconsistent: {message} | exception | error | rust, tool-calls, streaming, invariant, internal |
| duplicate tool name `{name}` | validation | error | rust, tools, validation, chat, duplicate |
| tool_choice requires at least one available tool | validation | error | rust, tool-choice, validation, chat |
| tool_choice function `{name}` was not found in the available | validation | error | rust, tool-choice, validation, typo, chat |
| failed to build structural tag: {message} | exception | error | rust, structural-tag, guided-decoding, grammar, validation |
| failed to render jinja template | exception | error | rust, jinja, chat-template, renderer, minijinja |
| failed to read chat template file | exception | error | rust, io, chat-template, file-permissions, configuration |
| chat template looks like a file path but does not exist | exception | error | rust, chat-template, file-not-found, configuration, docker |
| failed to parse chat_template.json | exception | error | rust, json, chat-template, corruption, download |
| chat_template.json does not contain a valid template | exception | error | rust, chat-template, json, validation, transformers |
| tool response messages require a tool_call_id; use ChatMessa | panic | critical | rust, panic, chat-message, tool-response, api-misuse |
| cannot combine `--headless` with `--data-parallel-size-local | validation | error | rust, cli, configuration, headless, argument-conflict |
| OpenAI server shut down unexpectedly without error | exception | error | rust, server, shutdown, lifecycle, supervisor |
| managed Python headless engine exited unexpectedly with stat | exception | critical | rust, engine, process-exit, oom, cuda, supervisor |
| cannot use in-process coordinator with bootstrapped transpor | panic | critical | rust, config, coordinator, transport, startup |
| messagepack encode failed for {target_type}: {message} | exception | error | rust, serialization, messagepack, protocol |
| messagepack decode failed for {target_type}: {message} | exception | error | rust, deserialization, messagepack, version-skew, protocol |
| messagepack value decode failed | exception | error | rust, deserialization, messagepack, corruption |
| messagepack ext value decode failed: {message} | exception | error | rust, messagepack, logprobs, deserialization, version-skew |
| invalid structured outputs params: {message} | validation | error | rust, structured-outputs, validation, request-building |
| io error | exception | error | rust, io, error-handling, diagnostics |
| transport error | exception | error | rust, zeromq, network, transport |
| ZMQ runtime task failed | exception | error | rust, tokio, zeromq, tasks, panic |
| engine core reported fatal failure | exception | critical | rust, engine-crash, lifecycle, zeromq |
| startup handshake timed out while waiting for {stage} after | exception | error | rust, startup, handshake, timeout, configuration |
| engine input registration timed out after {timeout:?} | exception | error | rust, startup, registration, timeout, network |
| unexpected engine id in startup handshake: expected {expecte | exception | error | rust, handshake, identity, configuration, duplicate |
| unexpected startup handshake message: {message} | exception | error | rust, handshake, protocol, validation, version-skew |
| unexpected non-control output on coordinator path: {message} | exception | error | rust, coordinator, protocol, routing |
| unexpected output on main dispatcher path: {message} | exception | error | rust, dispatcher, output-routing, version-skew, protocol |
| coordinator requires a Python-compatible two-byte engine id, | validation | error | rust, coordinator, identity, configuration |
| unsupported auxiliary frame(s): expected 1 frame, got {frame | exception | error | rust, zeromq, multimodal, aux-frames, limits |
| external coordinator mode is not implemented yet | exception | error | rust, coordinator, not-implemented, configuration |
| unsupported field `{field}` in {context} | validation | error | rust, validation, unsupported-feature, request-building |
| engine control channel closed unexpectedly: {message} | exception | critical | zmq, engine-core, ipc, process-crash, rust |
| request `{request_id}` is already in flight | validation | error | request-id, duplicate, api-misuse, rust |
| data parallel rank {rank} is not connected to this frontend; | validation | error | data-parallel, configuration, engine-core, rust |
| engine-core output dispatcher closed: {message} | exception | critical | zmq, dispatcher, engine-core, process-crash, rust |
| engine-core client is closed: {message} | exception | error | lifecycle, shutdown, api-misuse, rust |
| request output stream for `{request_id}` closed unexpectedly | exception | error | streaming, request-lifecycle, engine-core, rust |
| utility call `{method}` failed (call_id={call_id}): {message | exception | error | rpc, utility-call, engine-core, rust |
| utility call `{method}` returned an invalid result (call_id= | exception | error | serialization, version-skew, utility-call, rust |
| utility call `{method}` closed unexpectedly (call_id={call_i | exception | error | rpc, shutdown-race, utility-call, rust |
| utility call `{method}` returned inconsistent results across | exception | warning | data-parallel, utility-call, consistency, rust |
| failed to build vLLM ZMQ runtime | panic | critical | panic, tokio, resource-exhaustion, zmq, rust |
| generate request `{request_id}` has an empty prompt_token_id | exception | error | validation, prompt, api-misuse, rust |
| engine-core error | exception | error | wrapper, engine-core, error-chaining, rust |
| tokenizer is missing reasoning delimiter token `{token}` | exception | error | tokenizer, reasoning-parser, configuration, rust |
| `{name}` only provides a unified parser; the same reasoning | exception | error | reasoning-parser, tool-parser, unified-parser, configuration, rust |
| tool parser parsing failed: {message} | exception | warning | tool-parser, streaming, model-output, rust |
| `{name}` only provides a unified parser; the same reasoning | exception | error | tool-parser, unified-parser, configuration, rust |
| combined parser is constructed from split parser instances | exception | error | unified-parser, api-misuse, constructor, rust |
| tokenizer is missing unified parser token `{token}` | exception | error | tokenizer, unified-parser, configuration, rust |
| invalid --allowed-origins value {origin:?}: {e} | validation | error | cors, configuration, http-header, server, rust |
| invalid --allowed-methods value {method:?}: {e} | validation | error | configuration, cors, http, rust, vllm, startup |
| invalid --allowed-headers value {header:?}: {e} | validation | error | configuration, cors, http-headers, rust, vllm, startup |
| --ssl-certfile is required to enable TLS; --ssl-keyfile/--ss | validation | error | configuration, tls, ssl, security, rust, vllm, startup |
| --ssl-cert-reqs must be 0 (none), 1 (optional), or 2 (requir | validation | error | configuration, tls, ssl, security, rust, vllm, startup |
| --ssl-ca-certs is required when --ssl-cert-reqs is {} (clien | validation | error | configuration, tls, mtls, security, rust, vllm, startup |
| max_logprobs must be non-negative or -1, got {} | validation | error | configuration, logprobs, validation, rust, vllm, startup |
| data parallel size must be at least 1 | validation | error | configuration, data-parallel, validation, rust, vllm, startup |
| data parallel size ({}) exceeds the two-byte engine identity | validation | error | configuration, data-parallel, limits, rust, vllm, startup |
| managed frontend engine count ({engine_count}) must equal da | validation | error | configuration, data-parallel, transport, handshake, rust, vllm, startup |
| engine count must be at least 1 | validation | error | configuration, transport, bootstrap, data-parallel, rust, vllm, startup |
| engine start index does not fit usize | validation | error | configuration, transport, bootstrap, validation, rust, vllm, startup |
| engine start index + engine count overflows | validation | error | configuration, transport, bootstrap, overflow, rust, vllm, startup |
| connected engine range [{engine_start_index}, {engine_end_in | validation | error | configuration, transport, bootstrap, data-parallel, rust, vllm, startup |
| max_logprobs must be non-negative or -1 | validation | error | configuration, logprobs, render, validation, rust, vllm, startup |
| failed to build request runtime | panic | critical | panic, tokio, runtime, resources, limits, rust, vllm, startup |
| server info value must serialize | panic | critical | panic, serialization, serde, server-info, rust, vllm |
| tokenizer error: {0} | exception | error | tokenizer, huggingface, model-loading, network, rust, vllm |
| text request `{request_id}` must contain at least one prompt | validation | error | request-validation, prompt, token-ids, api, rust, vllm |
| text request `{request_id}` stop strings cannot be empty | validation | error | request-validation, stop-strings, sampling, api, rust, vllm |
| this model's maximum context length is {max_model_len} token | validation | error | request-validation, context-length, prompt, tokenizer, api, rust, vllm |
| `min_tokens` must be less than or equal to `max_tokens`, got | validation | error | rust, validation, sampling-params, vllm |
| `thinking_token_budget` must be a non-negative integer or -1 | validation | error | rust, validation, thinking, sampling-params, vllm |
| invalid repetition detection params: {message} | validation | error | rust, validation, repetition-detection, vllm |
| text request stream `{request_id}` closed before terminal ou | exception | error | rust, streaming, engine-core, vllm |
| {parameter} must be non-negative or -1, got {value} | validation | error | rust, validation, logprobs, vllm |
| requested {parameter} of {requested}, which is greater than | validation | error | rust, validation, logprobs, limits, vllm |
| requested logprob_token_ids of length {requested}, which is | validation | error | rust, validation, logprobs, vllm |
| when both logprobs and logprob_token_ids are set, logprobs m | validation | error | rust, validation, logprobs, vllm |
| {parameter} must be a finite number, got {value} | validation | error | rust, validation, sampling-params, vllm |
| {parameter} must be in {expected}, got {value} | validation | error | rust, validation, sampling-params, vllm |
| allowed_token_ids should not be empty | validation | error | rust, validation, token-ids, vllm |
| token_id(s) {token_ids:?} in {parameter} are out of vocabula | validation | error | rust, validation, token-ids, tokenizer, vllm |
| tokenizer error: {0} | exception | error | rust, tokenizer, huggingface, vllm |
| failed to parse `RUST_LOG` | panic | critical | rust, tracing, environment, startup, vllm |
| Output structure mismatch: {actual_spec} != {expected_spec} | exception | error | python, benchmark, helion, kernels, testing |
| Numerics check failed for case {case}:\n{e} | exception | error | python, benchmark, helion, numerics, testing |
| Cannot find CMake executable | exception | error | python, build, cmake, setup, vllm |
| No compatible wheel found for {arch} at {simple_url} | exception | error | python, packaging, wheels, network, vllm |
| Wheel metadata missing path: {wheel} | exception | error | python, packaging, wheels, metadata, vllm |
| No precompiled vllm wheel found for architecture {arch} from | exception | error | build, wheels, precompiled, architecture, setup-py |
| Unknown runtime environment | exception | error | build, environment-detection, setup-py, platform |
| Unsupported platform, please use CUDA, ROCm, or CPU. | exception | error | build, requirements, platform, setup-py |
| Could not determine Python executable. Please provide it man | exception | error | tooling, cmake, environment, interactive-prompt |
| CUDA is required for this proof. | exception | error | cuda, tooling, gpus, numerics |
| Failed to find the NIXL wheel after building it. | exception | error | nixl, build, wheels, ubuntu, tooling |
| Failed to find the repaired NIXL wheel. | exception | error | nixl, auditwheel, build, ubuntu, tooling |
| Unsupported file type: {file_path} | exception | warning | pre-commit, lint, spdx, license, tooling |
| ❌ line({node.lineno}): {message} | validation | warning | pre-commit, lint, config, ast, tooling |
| Unexpected level value ({args.level}) | exception | error | profiler, cli, argument-validation, tooling |
| PyYAML is required. Install it with: pip install pyyaml | exception | error | dependencies, pyyaml, cli, tooling |
| Interactive input is unavailable. Pass --model and --hardwar | exception | error | cli, interactive-prompt, ci, tooling |
| No selectable items found. | exception | error | cli, recipes, search, tooling |
| No recipe model matched {requested!r}. | exception | error | cli, recipes, search, tooling |
| Hardware {requested!r} is not available for this model. Avai | exception | error | cli, recipes, hardware, validation, tooling |
| Hardware recipe JSON does not contain a usable `strategy` fi | exception | error | cli, recipes, json, validation, tooling |
| Hardware recipe JSON `alternatives` must be an object when p | exception | error | cli, recipes, json, validation, tooling |
| Strategy {requested!r} is not available for this model/hardw | exception | error | cli, recipes, strategy, validation, tooling |
| {models_url} did not return a model list. | exception | error | network, recipes, api, json, tooling |
| Selected model {model.get('hf_id')!r} has no JSON API path. | exception | error | vllm-recipes, api-discovery, schema-validation |
| {model_json_url} did not return a JSON object. | exception | error | vllm-recipes, api-discovery, json-shape |
| Model {model.get('hf_id')!r} has no rendered recommended_com | exception | error | vllm-recipes, api-discovery, schema-validation |
| Model {model.get('hf_id')!r} has no per-hardware renderings | exception | error | vllm-recipes, api-discovery, hardware |
| Model {model.get('hf_id')!r} has no usable hardware JSON pat | exception | error | vllm-recipes, api-discovery, data-quality |
| {hardware_json_url} did not return a JSON object. | exception | error | vllm-recipes, api-discovery, json-shape |
| Cannot merge nested option {'.'.join(path)!r}: {part!r} is a | exception | error | vllm-recipes, argv-parsing, config-merge |
| Expected recipe argv to start with: ['vllm', 'serve', MODEL, | exception | error | vllm-recipes, argv-parsing, cli |
| Expected model after 'vllm serve', got {model!r} | exception | error | vllm-recipes, argv-parsing, cli |
| -O is missing its value | exception | error | vllm-recipes, argv-parsing, optimization-level |
| {token} is missing its value | exception | error | vllm-recipes, argv-parsing, parallelism |
| Unexpected positional/short argument {token!r}. The converte | exception | error | vllm-recipes, argv-parsing, cli, long-form-options |
| Recipe deploy_type={deploy_type!r} is not a single-node depl | exception | error | vllm-recipes, deploy-type, multi-node |
| This recipe is a multi-process deployment and cannot be repr | exception | error | vllm-recipes, multi-process, prefill-decode, deploy-type |
| Recipe JSON does not contain an `argv` field | exception | error | vllm-recipes, schema-drift, argv-parsing |
| Recipe `env` must be an object, got {type(env).__name__} | exception | error | vllm-recipes, env, schema-validation |
| Do not combine a positional recipe JSON source with --model/ | exception | error | vllm-recipes, cli-usage, argument-validation |
| Recipe JSON must be a JSON object. | exception | error | vllm-recipes, json-shape, input-validation |
| Invalid wheel filename format: {wheel_name} | exception | error | rocm, packaging, wheel, pep427 |
| module {__package__} has no attribute {name} | exception | error | vllm, lazy-import, pep562, attributeerror, api-surface |
| padded_n is not supported with TRTLLM 8x4 scale layout. | exception | error | quantization, nvfp4, trtllm, gpu, validation |
| asymmetric int8 activation quantization is unsupported on XP | exception | error | quantization, int8, xpu, intel-gpu, platform-support |
| The optimized moe_wna16_gemm kernel is only available on CUD | exception | error | moe, quantization, weight-only, cuda-only, platform-support |
| The fused grouped_topk kernel is only available on CUDA plat | exception | error | moe, topk, routing, cuda-only, platform-support |
| `a` must have at least 1 dimension. | exception | error | quantization, mx-format, fp4, validation, shape |
| last dim of `a` must be divisible by 32, got {a.size(-1)}. | exception | error | quantization, mx-format, fp4, validation, shape, alignment |
| `a` and `b` must be on the same device. | exception | error | quantization, mx-format, device-mismatch, validation, multi-gpu |
| The `_qutlass_C` extension is not loaded. Make sure your cus | exception | error | quantization, extension-loading, cuda, environment |
| invalid method {method!r}, must be 'quest' or 'abs_max' | exception | error | quantization, mx-format, validation, api-misuse |
| Could not open video file {path} | exception | error | video, multimodal, opencv, file-io, validation |
| Could not read enough frames from video file {path} (expecte | exception | error | video, multimodal, opencv, data-quality, validation |
| Could not collect pip list output (pip or uv module not avai | exception | warning | diagnostics, environment, pip, uv, packaging |
| Expected {ty} but got {type(value)} for {value} | exception | error | compilation, cache, version-skew, torch-compile |
| PostGradPassManager can not be kept in CompilationConfig. | exception | error | compilation, torch-compile, configuration, api-misuse |
| Nested BreakableCUDAGraphCapture is not supported. | exception | error | cuda-graph, compilation, runtime, nested-context |
| call_module is not allowed for codegen target {target}. | exception | error | compilation, torch-compile, fx-graph, codegen, internal |
| Unsupported node from codegen: {node.format_node()} | exception | error | compilation, torch-compile, fx-graph, codegen, internal |
| caching is not supported | exception | error | compilation, cache, custom-backend, api-contract |
| The compiled artifact is not serializable. This usually mean | exception | error | compilation, cache, serialization, torch-compile, version-skew |
| vLLM failed to compile the model. The most likely reason for | exception | error | compilation, torch-compile, cache, vllm |
| Attribute {key} not exists in the runnable of cudagraph wrap | exception | error | cudagraph, attribute-access, wrapper, vllm |
| decorated class should have a forward method. | exception | error | torch-compile, decorators, vllm, model-porting |
| No dynamic dimensions found in the forward method of {cls}. | exception | error | torch-compile, decorators, dynamic-shapes, vllm |
| Argument {k} not found in the forward method of {cls} | exception | error | torch-compile, decorators, dynamic-shapes, vllm |
| Source code has changed since the last compilation. Recompil | exception | error | compilation, cache-invalidation, source-tracking, vllm |
| {init} received a positional argument of type {arg_type}, bu | exception | error | torch-compile, decorators, type-checking, vllm |
| shape_id='{shape_id}' requires PyTorch >= 2.11.0 | exception | error | torch-compile, version-compat, dynamic-shapes, vllm |
| Unsupported dynamic dimensions {dims} for argument {k} with | exception | error | torch-compile, dynamic-shapes, type-validation, vllm |
| CUDA graph capturing detected at an inappropriate time. This | exception | error | cudagraph, runtime-monitor, cuda, vllm |
| Input {arg} to maybe_inplace node {node} is used again after | exception | error | compilation, inplace-ops, custom-ops, vllm |
| cudagraph_capture_sizes not supported in compile_sizes.This | exception | error | compilation, piecewise-backend, cudagraph, vllm |
| Compilation mode cannot be NO_COMPILATION | exception | error | torch-compile, configuration, vllm |
| aot_compile is not supported by the current configuration. P | exception | error | aot-compile, version-compat, torch-compile, vllm |
| Assigning / modifying buffers of nn.Module during forward pa | exception | error | cudagraph, bytecode-hook, silent-corruption, vllm |
| Unknown KV cache group kind '{kind}' in backend_per_kind. Va | validation | error | configuration, attention-backend, kv-cache, vllm |
| Invalid compilation mode: {value}. Valid modes are: {', '.jo | validation | error | configuration, compilation-mode, vllm |
| compile_cache_save_format must be 'binary' or 'unpacked', go | validation | error | configuration, compile-cache, vllm |
| use_inductor_graph_partition is only supported with torch>=2 | validation | error | configuration, version-compat, torch, vllm |
| Invalid syntax '{op}' for custom op, must be 'all', 'none', | validation | error | configuration, custom-ops, vllm |
| custom_ops can contain only one base mode: 'all' or 'none' | validation | error | |
| custom_ops cannot both enable and disable the same operation | validation | error | |
| Invalid backend for piecewise compilation: {self.backend} | validation | error | |
| encoder_cudagraph_max_vision_items_per_batch must be non-neg | validation | error | |
| encoder_cudagraph_max_frames_per_batch must be non-negative | validation | error | |
| All encoder_cudagraph_token_budgets must be positive, got {s | validation | error | |
| No compilation mode is set. This method should only be calle | validation | error | |
| No compilation mode is set. | validation | error | |
| custom_ops must contain exactly one base mode: 'all' or 'non | validation | error | |
| CUDAGraphMode.{cudagraph_mode.name} is not supported with {m | validation | error | |
| CUDAGraphMode.{cudagraph_mode.name} is not supported with {m | validation | error | |
| max_num_seqs ({max_num_reqs}) exceeds available Mamba cache | validation | error | |
| Can't determine cudagraph shapes that are both a multiple of | validation | error | |
| No valid cudagraph sizes after rounding to multiple of {mult | validation | error | |
| Failed to infer device type, please set the environment vari | validation | error | |
| manager_config requires encoder_cache_manager_cls to be set. | validation | error | |
| Unsupported ec_role: {self.ec_role}. Supported roles are {ge | validation | error | |
| Please specify ec_role when ec_connector is set, supported r | validation | error | |
| Unsupported kv_role: {self.kv_role}. Supported roles are {ge | validation | error | |
| Please specify kv_role when kv_connector is set, supported r | validation | error | |
| max_cpu_loras ({self.max_cpu_loras}) must be >= max_loras ({ | validation | error | |
| Dual CUDA streams are only supported on CUDA platforms. | validation | error | |
| Inkling LoRA requires INKLING_MULTIMEM_AR=0: the Lamport fus | validation | error | |
| Unknown Mamba SSU backend: '{name}'. Valid options are: {val | validation | error | |
| Unknown Mamba SSU algorithm: '{self.ssu_algorithm}'. Valid o | validation | error | |
| Mamba SSU algorithm selection is only supported with the Fla | validation | error | |
| Stochastic rounding for Mamba cache is only supported on NVI | validation | error | |
| Stochastic rounding for Mamba cache with triton backend requ | validation | error | |
| Sleep mode is not supported on current platform. | validation | error | |
| cumem allocator is not supported on current platform. | validation | error | |
| Embedding models do not support `--runner {self.runner_type} | validation | error | |
| This model does not support `--runner generate`. | validation | error | |
| This model does not support `--runner pooling`. You can pass | validation | error | |
| Video pruning method '{pruning_spec[0]}' is not supported by | validation | error | |
| Cannot use --renderer-num-workers > 1 with the multimodal pr | validation | error | |
| tokenizer must be a string, got {type(self.tokenizer).__name | validation | error | |
| max_model_len must be a positive integer, got {type(self.max | validation | error | |
| Quantization method {name} is an override but is has not bee | validation | error | |
| Quantization method specified in the model config ({quant_me | validation | error | |
| Unknown quantization method: {self.quantization}. Must be on | validation | error | |
| The quantization method %s is deprecated and will be removed | validation | error | quantization, deprecation, startup, config |
| Number of experts in the model must be greater than 0 when e | validation | error | parallelism, moe, config, startup |
| Total number of attention heads ({total_num_attention_heads} | validation | error | tensor-parallelism, attention, config, startup |
| Pipeline parallelism is not supported for this model. Suppor | validation | error | pipeline-parallelism, model-support, config, startup |
| Decode context parallelism for GQA/MQA requires `--tensor-pa | validation | error | parallelism, decode-context-parallel, gqa, config |
| `--decode-context-parallel-size` ({decode_context_parallel_s | validation | error | parallelism, decode-context-parallel, gqa, config |
| The model's number of query heads per KV head ({num_q_per_kv | validation | error | parallelism, decode-context-parallel, gqa, config |
| mm_tensor_ipc='torch_shm' is not supported with data_paralle | validation | error | multimodal, ipc, parallelism, config |
| The model is an hybrid without a layers_block_type or an att | validation | error | hybrid-model, hf-config, config, startup |
| The model is not multimodal. | validation | error | multimodal, api-misuse, config |
| Unsupported task: {pooling_task!r} Supported tasks: {support | exception | error | pooling, task-config, config |
| The model type {model_type!r} does not support float16. Reas | validation | error | dtype, float16, model-support, config |
| Unknown dtype: {dtype!r} | validation | error | dtype, config, validation |
| Unknown dtype: {dtype} | validation | error | dtype, type-error, validation |
| Unknown dtype: {head_dtype!r} | validation | error | dtype, hf-config, head, validation |
| Unknown dtype: {head_dtype} | validation | error | dtype, hf-config, type-error |
| {msg} To allow overriding this maximum, set the env var VLLM | validation | error | max-model-len, context-length, config, startup |
| a model must have at least one layer | validation | error | model-arch, internal-api, validation |
| {f.name!r} varies across layers and has no whole-model value | validation | error | model-arch, heterogeneous-layers, internal-api |
| got {len(layers)} per-layer configs for a model with {merged | validation | error | model-arch, heterogeneous-layers, checkpoint, internal-api |
| Attention backend 'XFORMERS' has been removed (See PR #29262 | validation | error | vllm, config, attention-backend, multimodal, migration |
| 'mm_shm_cache_max_object_size_mb' should only be set when 'm | validation | error | vllm, config, multimodal, cache, validation |
| 'mm_encoder_fp8_scale_path' and 'mm_encoder_fp8_scale_save_p | validation | error | vllm, config, multimodal, fp8, quantization, validation |
| 'mm_encoder_fp8_scale_save_path' cannot be used with 'mm_enc | validation | error | vllm, config, multimodal, fp8, mutually-exclusive, validation |
| FP8 scale file not found: {scale_path} | exception | error | vllm, config, multimodal, fp8, file-not-found, paths |
| Parent directory for FP8 scale save path not found: {save_pa | exception | error | vllm, config, multimodal, fp8, file-not-found, paths |
| Invalid "device" in mm_processor_kwargs: {device!r}. Expecte | validation | error | vllm, config, multimodal, device, torch, validation |
| Cannot run the multi-modal processor on {device_type!r}: thi | validation | error | vllm, config, multimodal, device, gpu, oom, disaggregation |
| OpenTelemetry is not available. Unable to configure 'otlp_tr | validation | error | vllm, config, observability, opentelemetry, dependencies |
| collect_detailed_traces requires `--otlp-traces-endpoint` to | validation | error | vllm, config, observability, tracing, validation |
| offload_num_in_group ({self.prefetch.offload_num_in_group}) | validation | error | vllm, config, offloading, kv-cache, validation |
| offload_prefetch_step ({self.prefetch.offload_prefetch_step} | validation | error | vllm, config, offloading, prefetch, validation |
| Async EPLB is only supported with the default policy. | validation | error | vllm, config, eplb, moe, load-balancing, validation |
| {self.communicator} communicator is incompatible with async | validation | error | vllm, config, eplb, nccl, moe, validation |
| log_balancedness_interval must be greater than 0. | validation | error | vllm, config, eplb, logging, validation |
| numa_bind_nodes must not be empty. | validation | error | vllm, config, numa, parallel, validation |
| numa_bind_nodes must contain non-negative integers. | validation | error | vllm, config, numa, parallel, validation |
| numa_bind_cpus must not be empty. | validation | error | vllm, config, numa, cpuset, validation |
| numa_bind_cpus entries must not be empty. | validation | error | vllm, config, numa, cpuset, parsing, validation |
| numa_bind_cpus entries must use numactl CPU list syntax, for | validation | error | vllm, config, numa, cpuset, regex, validation |
| numa_bind_cpus ranges must be ascending, but got '{cpuset}'. | validation | error | numa, cpu-affinity, configuration, parallelism |
| Invalid value of `_api_process_rank`. Expected to be `-1` or | validation | error | api-server, configuration, internal, parallelism |
| Fault tolerance requires a single API server process (--api- | validation | error | fault-tolerance, api-server, configuration, parallelism |
| data_parallel_size_local ({self.data_parallel_size_local}) m | validation | error | data-parallel, configuration, multi-node, parallelism |
| data_parallel_external_lb can only be set when data_parallel | validation | error | data-parallel, load-balancing, configuration, parallelism |
| numa_bind_nodes and numa_bind_cpus require numa_bind=True. | validation | error | numa, cpu-affinity, configuration |
| Expert parallelism load balancing is only supported on CUDA | validation | error | eplb, moe, platform-support, configuration |
| enable_expert_parallel must be True to use EPLB. | validation | error | eplb, expert-parallel, moe, configuration |
| EPLB requires tensor, prefill-context, or data parallelism, | validation | error | eplb, expert-parallel, moe, parallelism, configuration |
| num_redundant_experts is set to {self.eplb_config.num_redund | validation | error | eplb, moe, configuration |
| PCP does not support data parallelism yet. | validation | error | context-parallelism, data-parallel, parallelism, configuration |
| tp_size={tp} must be divisible by dcp_size={dcp}. | validation | error | context-parallelism, tensor-parallel, parallelism, configuration |
| When PCP is enabled, DCP must be disabled, span the PCP axis | validation | error | context-parallelism, parallelism, configuration |
| dcp_comm_backend='a2a' requires decode_context_parallel_size | validation | error | context-parallelism, communication-backend, configuration |
| Elastic EP is only supported with enable_eplb=True. | validation | error | elastic-ep, eplb, moe, configuration |
| Elastic EP is not supported with pipeline parallelism (pipel | validation | error | elastic-ep, pipeline-parallel, moe, configuration |
| Elastic EP is not compatible with data_parallel_external_lb | validation | error | elastic-ep, load-balancing, data-parallel, not-implemented, configuration |
| Elastic EP with async EPLB requires the NIXL package. Either | validation | error | elastic-ep, eplb, nixl, dependencies, configuration |
| data_parallel_rank ({self.data_parallel_rank}) must be in th | validation | error | data-parallel, multi-node, launcher, configuration |
| Offline data parallel mode is not supported/useful for dense | validation | error | offline-inference, data-parallel, environment-variables, dense-model |
| World size ({self.world_size}) is larger than the number of | validation | critical | parallelism, gpu, configuration, startup |
| nnodes > 1 can only be set when distributed executor backend | validation | critical | parallelism, multi-node, ray, configuration |
| Unrecognized distributed executor backend {self.distributed_ | validation | error | parallelism, executor, type-validation, api-misuse |
| Unable to use nsight profiling unless workers run with Ray. | validation | error | profiling, nsight, ray, configuration |
| Parameter `normalize` was removed; use `use_activation` inst | validation | error | pooling, reward-model, migration, removed-parameter |
| logit_sigma cannot be 0 (division by zero) | validation | error | pooling, classification, calibration, configuration |
| Cannot set both `pooling_type` and `seq_pooling_type` | validation | error | pooling, configuration, conflicting-fields |
| Cannot set both `pooling_type` and `tok_pooling_type` | validation | error | pooling, configuration, conflicting-fields |
| {pooling_type} | validation | error | pooling, invalid-value, configuration |
| seq_pooling_type is not set; it should be resolved by ModelC | validation | error | pooling, internal-contract, api-misuse |
| tok_pooling_type is not set; it should be resolved by ModelC | validation | error | pooling, internal-contract, api-misuse |
| torch_profiler_dir is only applicable when profiler is set t | validation | error | profiling, torch-profiler, configuration |
| torch_profiler_dir must be set when profiler is 'torch' | validation | error | profiling, torch-profiler, configuration |
| {options} only applicable when profiler is set to 'proton' | validation | error | profiling, proton, configuration |
| proton_profiler_dir must be set when profiler is 'proton' | validation | error | profiling, proton, configuration |
| proton_profiler_dir must be a local directory | validation | error | profiling, proton, path, configuration |
| chrome_trace output requires proton_data='trace' | validation | error | profiling, proton, configuration |
| {output_format} output requires proton_data='tree' | validation | error | profiling, proton, configuration |
| capture_torch_profiler is only applicable when profiler is s | validation | error | profiling, cuda-graph, configuration |
| expected str or QuantKey, got {type(v).__name__} | validation | error | quantization, type-validation, pydantic, api-misuse |
| unknown quantization name {v!r}; expected one of {sorted(QUA | validation | error | quantization, config, validation, pydantic |
| online shorthand {v!r} does not define a {field_name} spec | validation | error | quantization, config, validation |
| quantization_config is only supported when quantization is o | validation | error | quantization, cli, config, validation |
| ReasoningConfig: failed to tokenize reasoning strings: reaso | exception | error | reasoning, tokenizer, config, validation |
| max_num_batched_tokens ({self.max_num_batched_tokens}) is sm | exception | error | scheduler, config, context-length, validation |
| max_num_batched_tokens ({self.max_num_batched_tokens}) must | exception | error | scheduler, config, validation |
| long_prefill_token_threshold ({self.long_prefill_token_thres | exception | error | scheduler, config, validation |
| rejection_sample_method='synthetic' requires exactly one of | exception | error | speculative-decoding, config, validation |
| synthetic_acceptance_rates must have length {n}, got {rates} | exception | error | speculative-decoding, config, validation |
| synthetic_acceptance_rates entries must be in [0, 1], got {r | exception | error | speculative-decoding, config, validation, math |
| synthetic_acceptance_rates must be non-increasing, got {rate | exception | error | speculative-decoding, config, validation, math |
| synthetic_acceptance_length must be in [1, {n + 1}], got {le | exception | error | speculative-decoding, config, validation, math |
| mtp_layer_types must have one entry per MTP layer: got {len( | exception | error | speculative-decoding, mtp, checkpoint, config |
| The Inkling checkpoint does not contain MTP weights | exception | error | speculative-decoding, mtp, checkpoint, config |
| target_model_config must be present for mtp | exception | error | speculative-decoding, mtp, config, api-misuse |
| target_model_config must be present for dspark | exception | error | speculative-decoding, dspark, config, api-misuse |
| method='custom_class' requires 'model' to contain the custom | exception | error | speculative-decoding, custom-class, config, api-misuse |
| num_speculative_tokens was provided but without speculative | exception | error | speculative-decoding, config, cli, api-misuse |
| Either prompt_lookup_max or prompt_lookup_min must be provid | exception | error | speculative-decoding, ngram, config, dead-code |
| prompt_lookup_min={self.prompt_lookup_min} must be <= prompt | exception | error | speculative-decoding, ngram, config, validation |
| Unsupported speculative method: '{self.method}' | exception | error | speculative-decoding, config, not-implemented, method-dispatch |
| MLA DSpark does not currently support decode context paralle | exception | error | speculative-decoding, dspark, context-parallelism, config |
| num_speculative_tokens:{self.num_speculative_tokens} must be | exception | error | speculative-decoding, mtp, num-speculative-tokens, validation |
| A speculative model was provided, but `num_speculative_token | exception | error | speculative-decoding, config, num-speculative-tokens, required-field |
| dspark_draft_topk is only supported by DSpark | exception | error | speculative-decoding, dspark, config, invalid-combination |
| dspark_draft_topk must be between 1 and the draft vocabulary | exception | error | speculative-decoding, dspark, topk, range-validation |
| dspark_draft_topk is only supported by Qwen3DSparkModel | exception | error | speculative-decoding, dspark, topk, architecture-check |
| Adaptive verification only supported with DSpark | exception | error | speculative-decoding, dspark, adaptive-verification, invalid-combination |
| Arctic Inference is required for suffix decoding. Install vi | exception | error | speculative-decoding, suffix-decoding, dependency, import-error |
| suffix_decoding_max_tree_depth={self.suffix_decoding_max_tre | exception | error | speculative-decoding, suffix-decoding, range-validation, config |
| suffix_decoding_max_cached_requests={self.suffix_decoding_ma | exception | error | speculative-decoding, suffix-decoding, cache, range-validation |
| suffix_decoding_max_spec_factor={self.suffix_decoding_max_sp | exception | error | speculative-decoding, suffix-decoding, range-validation, config |
| suffix_decoding_min_token_prob={self.suffix_decoding_min_tok | exception | error | speculative-decoding, suffix-decoding, probability, range-validation |
| {speculative_max_model_len=} cannot be larger than {draft_ma | exception | error | speculative-decoding, max-model-len, draft-model, validation |
| {speculative_max_model_len=} cannot be larger than {target_m | exception | error | speculative-decoding, max-model-len, target-model, validation |
| {speculative_draft_tensor_parallel_size=} cannot be other va | exception | error | speculative-decoding, tensor-parallelism, draft-model, config |
| 'tensor_parallel_size' is not a valid argument in the specul | exception | error | speculative-decoding, config, tensor-parallelism, renamed-field |
| num_speculative_tokens must be provided with speculative mod | exception | error | speculative-decoding, num-speculative-tokens, required-field, pydantic |
| Expected num_speculative_tokens to be greater than zero ({se | exception | error | speculative-decoding, num-speculative-tokens, range-validation, pydantic |
| synthetic_acceptance_rates / synthetic_acceptance_length are | exception | error | speculative-decoding, config, validation |
| use_heterogeneous_vocab only works with method='draft_model' | exception | error | speculative-decoding, config, tokenizer, validation |
| use_heterogeneous_vocab currently only supports greedy draft | exception | error | speculative-decoding, config, sampling, validation |
| Target and draft model should have the same vocabulary size. | validation | error | speculative-decoding, tokenizer, config, validation |
| disable_any_whitespace is only supported for xgrammar and gu | validation | error | structured-output, grammar, config, validation |
| disable_additional_properties is only supported for the guid | validation | error | structured-output, grammar, json-schema, config |
| {field_path} is not a valid config field | validation | error | config, validation, typo, api-misuse |
| Override for {field_path} must be a mapping or {expected_typ | validation | error | config, validation, type-mismatch, api-misuse |
| DeepSeek V4 does not support PIECEWISE CUDA graphs with Mode | validation | error | cuda-graph, deepseek, model-runner, config |
| The quantization method {model_config.quantization} is not s | validation | error | quantization, gpu-capability, hardware, config |
| {model_config.dtype} is not supported for quantization metho | validation | error | quantization, dtype, config, validation |
| KV connector {self.kv_transfer_config.kv_connector} is incom | validation | error | kv-transfer, allocator, environment, disaggregation |
| sampling distribution replay requires Model Runner V2 | validation | error | sampling, model-runner, config, validation |
| sampling distribution replay does not support speculative de | validation | error | sampling, speculative-decoding, config, incompatibility |
| sampling distribution replay does not support diffusion mode | validation | error | sampling, diffusion, config, incompatibility |
| sampling distribution replay does not support custom logits | validation | error | sampling, logits-processors, config, incompatibility |
| sampling distribution replay requires logprobs_mode='process | validation | error | sampling, logprobs, config, validation |
| --enable-return-routed-experts is incompatible with pipeline | validation | error | moe, pipeline-parallelism, expert-routing, config |
| --enable-return-routed-experts is incompatible with context | validation | error | moe, context-parallelism, expert-routing, config |
| --enable-return-routed-experts is incompatible with KV conne | validation | error | moe, expert-routing, kv-transfer, disaggregation |
| Stochastic rounding for Mamba cache requires the SSM cache t | validation | error | mamba, cache-dtype, stochastic-rounding, startup-config |
| Async scheduling is not compatible with ROCm DeepEP high-thr | validation | error | rocm, deepep, async-scheduling, dbo, moe |
| Currently, async scheduling is only supported with EAGLE/MTP | validation | error | async-scheduling, speculative-decoding, startup-config |
| Async scheduling is not compatible with disable_padded_draft | validation | error | async-scheduling, speculative-decoding, startup-config |
| `{executor_backend}` does not support async scheduling yet. | validation | error | async-scheduling, executor-backend, distributed |
| torch_shm is known to fail without VLLM_WORKER_MULTIPROC_MET | validation | error | multimodal, shared-memory, environment, multiprocessing |
| The Proton profiler currently supports NVIDIA CUDA only | validation | error | profiling, proton, platform-support, cuda |
| The Proton profiler requires CUDA graphs to be disabled. Use | validation | error | profiling, proton, cuda-graphs, startup-config |
| Fast prefill optimization for KV sharing is not compatible w | validation | error | kv-cache, kv-sharing, eagle, speculative-decoding |
| Prefill context parallelism requires Model Runner V2. Remove | validation | error | context-parallelism, model-runner, environment |
| Hybrid KV cache manager was explicitly enabled but is not su | validation | error | kv-cache, hybrid-attention, kv-connector, startup-config |
| max_num_scheduled_tokens is set to {self.scheduler_config.ma | validation | error | scheduler, speculative-decoding, batched-tokens |
| VllmConfig does not have enough slots to schedule a token an | validation | error | scheduler, speculative-decoding, batched-tokens |
| customized max_cudagraph_capture_size(={self.compilation_con | validation | error | cuda-graphs, compilation-config, startup-config |
| To load a model from object storage (S3/GCS/Azure), 'load_fo | validation | error | model-loading, object-storage, s3, load-format |
| Model Runner V2 requires Triton. | validation | error | triton, model-runner, environment, dependencies |
| Model Runner V2 does not yet support: {', '.join(unsupported | validation | error | model-runner, feature-support, startup-config |
| nvfp4 KV cache is not supported with MLA (Multi-head Latent | validation | error | kv-cache, nvfp4, mla, deepseek, quantization |
| --mamba-block-size can only be set with --enable-prefix-cach | validation | error | mamba, prefix-caching, block-size, startup-config |
| --use-replayssm is only supported for Nemotron-H models (got | validation | error | mamba, replayssm, nemotron, model-support |
| --use-replayssm supports prefix caching only in align mode; | validation | error | vllm, config, mamba, replayssm, prefix-caching |
| --use-replayssm does not support speculative decoding | validation | error | vllm, config, replayssm, speculative-decoding, mamba |
| --use-replayssm requires --mamba-backend triton | validation | error | vllm, config, mamba, triton, replayssm |
| --use-replayssm is incompatible with KV connectors (P/D disa | validation | error | vllm, config, replayssm, kv-transfer, disaggregation, mamba |
| Invalid HTTP URL: A valid HTTP URL must have scheme 'http' o | validation | error | vllm, url, validation, network, config |
| Sleep mode allocator is not available on platform {type(curr | exception | error | vllm, sleep-mode, platform, allocator, cuda, xpu |
| Sleep-mode backend '{name}' is already registered. | validation | error | vllm, sleep-mode, plugin, registry, duplicate |
| Unsupported sleep-mode backend '{name}'. Registered backends | validation | error | vllm, sleep-mode, config, registry, typo |
| Sleep-mode backend '{name}' is not supported on this platfor | validation | error | vllm, sleep-mode, platform, hardware, driver |
| torch.xpu.memory is not available | exception | error | vllm, xpu, torch, sleep-mode, environment |
| xpumem allocator extension is not available | exception | error | vllm, xpu, extension, vllm-xpu-kernels, sleep-mode |
| torch.xpu.memory.XPUPluggableAllocator is not available | exception | error | vllm, xpu, torch, version-mismatch, allocator |
| torch.xpu.memory MemPool APIs are not available (need MemPoo | exception | error | vllm, xpu, torch, mempool, version-mismatch |
| mori currently only support arch gfx942 and gfx950 | exception | error | vllm, rocm, mori, expert-parallel, gpu-arch |
| DeepEPv2 communicator properties query failed; networking ca | exception | critical | vllm, deepep, nccl, networking, distributed |
| DeepEPv2 requires NCCL GIN (GPU-Initiated Networking). This | exception | critical | vllm, deepep, nccl, ibgda, infiniband, distributed |
| Error happened when batch testing peer-to-peer access from { | exception | error | vllm, cuda, p2p, multi-gpu, diagnostics |
| Unknown all2all backend: {self.all2all_backend} | validation | error | vllm, distributed, all2all, config, typo |
| CUDART error: {error_str} | exception | error | cuda, distributed, native-bindings, runtime |
| Flashinfer allreduce is not supported for multi-node allredu | validation | error | distributed, flashinfer, allreduce, multi-node, config |
| Flashinfer allreduce quantization fusion is not supported fo | validation | error | distributed, flashinfer, allreduce, quantization, multi-node |
| Unsupported dtype {dtype}: should be one of int8, uint8, int | validation | error | nccl, dtype, distributed, validation |
| Unsupported op: {op} | validation | error | nccl, reduce-op, distributed, validation |
| NCCL error: {error_str} | exception | critical | nccl, distributed, network, multi-node |
| Expected 128 bytes for ncclUniqueId, got {len(data)} bytes | validation | error | nccl, serialization, distributed, validation |
| VLLM_ROCM_QUICK_REDUCE_MIN_SIZE_BYTES_MB must be non-negativ | validation | error | rocm, allreduce, env-var, config |
| VLLM_ROCM_QUICK_REDUCE_MIN_SIZE_BYTES_MB must be less than o | validation | error | rocm, allreduce, env-var, config |
| VLLM_ROCM_QUICK_REDUCE_QUANTIZATION_MIN_SIZE_KB must be non- | validation | error | rocm, allreduce, quantization, env-var |
| use_communication_streams is not supported | exception | error | ray, pipeline-parallel, distributed, not-implemented |
| cuda_stream other than the current stream is not supported | validation | error | ray, cuda-stream, pipeline-parallel, distributed |
| Actor {actor} not found in communicator group | validation | error | ray, pipeline-parallel, distributed, actor-management |
| RayPPCommunicator has been destroyed. | exception | critical | ray, pipeline-parallel, lifecycle, distributed |
| allgather is not supported | exception | error | ray, pipeline-parallel, collective, not-implemented |
| allreduce is not supported | exception | error | ray, pipeline-parallel, allreduce, not-implemented |
| reducescatter is not supported | exception | error | ray, pipeline-parallel, collective, not-implemented |
| Insufficient space in {shm_path}: {required_bytes / mib:.0f} | exception | error | shared-memory, docker, distributed, resource-limits |
| Only readers can dequeue | exception | error | distributed, shared-memory, api-misuse |
| Not enough space in the data buffer, try calling free_buf() | exception | error | shared-memory, out-of-memory, distributed |
| Unsupported object type '{type_name}' in metadata | validation | error | serialization, version-mismatch, shared-memory |
| Lock must be provided for readers. | validation | error | configuration, api-misuse, shared-memory |
| Unsupported data type for serialization: {type(data)} | validation | error | serialization, api-misuse, type-error |
| Key '{key}' already exists in the storage. | validation | error | api-misuse, key-conflict, shared-memory |
| Serialized object size ({buffer_size} bytes) exceeds max obj | validation | error | configuration, size-limit, serialization |
| Data for address:id '{address}:{monotonic_id}' has been modi | validation | error | shared-memory, stale-handle, race-condition |
| only dim 0 all-gatherv is supported | exception | error | xpu, distributed, unsupported-operation |
| ec_transfer_config must be set for ECConnectorBase | validation | error | configuration, ec-transfer, api-misuse |
| ec_cpu_bytes must be specified in ec_connector_extra_config | validation | error | configuration, ec-transfer, setup |
| Unknown ECConnectorRole: {role} | validation | error | api-misuse, ec-transfer, enum |
| Timed out waiting for EC mmap file to reach {expected_size} | exception | critical | ec-transfer, timeout, filesystem, startup |
| ec_transfer_config must be set for ECConnectorBase | validation | error | configuration, ec-transfer, api-misuse |
| Connector '{name}' is already registered. | validation | error | ec-transfer, plugin-registry, duplicate-registration |
| ec_transfer_config must be set to create a connector | validation | error | configuration, ec-transfer, factory |
| EC connect must not be None | validation | error | configuration, ec-transfer, factory |
| Unsupported connector type: {connector_name} | validation | error | configuration, ec-transfer, factory, unknown-class |
| No device communicator found | validation | error | elastic-ep, distributed, initialization |
| Worker has been garbage collected | exception | critical | elastic-ep, lifetime-management, garbage-collection, race-condition |
| Unknown execute method: {execute_method} | validation | error | |
| Another Elastic EP async method is active | exception | error | |
| DBO is not yet supported in elastic EP | exception | error | |
| NIXL is unavailable. | exception | error | |
| NIXL EPLB init failed: {name} | exception | error | |
| NIXL EPLB metadata key mismatch with rank {peer}: local={sor | exception | error | |
| NIXL EPLB nbytes_per_expert mismatch for {key} with rank {pe | exception | error | |
| NIXL transfer failed with state={state} | exception | error | |
| EPLB communicator 'pynccl' supports only cuda-like devices ( | exception | error | |
| EPLB communicator 'pynccl' requested but expert weights cont | exception | error | |
| EPLB communicator 'pynccl' requested but unavailable. | exception | error | |
| Failed to initialize PyNcclEplbCommunicator ({exc}). | exception | error | |
| Elastic EP requires 'torch_nccl', 'pynccl', or 'nixl' EPLB c | validation | error | |
| EPLB communicator 'nixl' requested but NIXL is unavailable. | exception | error | |
| EPLB communicator 'nixl' supports only cuda-like devices (go | exception | error | |
| Failed to initialize NixlEplbCommunicator ({exc}). | exception | error | |
| Unknown EPLB communicator backend: {backend} | validation | error | |
| Model: {} with config {} {} {} {} {} mismatch with new model | exception | error | |
| Invalid endpoint: must contain 'inproc' or 'tcp' | validation | error | |
| Unknown event publisher '{kind}' | validation | error | |
| kv_transfer_config must be set to create a connector | validation | error | kv-transfer, config, vllm, connector-factory |
| Connector {connector_cls.__name__} does not support HMA but | validation | error | kv-transfer, hma, config, vllm |
| Connector '{connector_name}' is not registered. | validation | error | kv-transfer, registry, config, vllm |
| Connector name is not set in KVTransferConfig | validation | error | kv-transfer, config, vllm |
| kv_connector_module_path cannot be an empty string. | validation | error | kv-transfer, config, validation, vllm |
| Class {connector_name} not found in {connector_module_path} | exception | error | kv-transfer, import, external-connector, vllm |
| Connector {connector_cls.__name__} uses deprecated 2-argumen | validation | error | kv-transfer, api-change, external-connector, vllm |
| Unsupported connector type: {connector_name} | validation | error | kv-transfer, registry, config, vllm |
| kv_transfer_config must be set for KVConnectorBase_V1 | validation | error | kv-transfer, config, base-class, vllm |
| {type(self).__name__} received pp_rank > 0 handshake metadat | validation | error | kv-transfer, pipeline-parallelism, handshake, vllm |
| Could not uniquely identify the extract-hidden-states KV cac | exception | error | kv-transfer, hidden-states, kv-cache-groups, vllm |
| Hidden-states block-size mismatch: derived {self._block_size | exception | error | kv-transfer, block-size, hybrid-kv-cache, vllm |
| FlexKV is not installed. Please install it to use FlexKVConn | exception | error | kv-transfer, import, flexkv, dependencies, vllm |
| hf3fs_fuse.io is not available. Please install the hf3fs_fus | exception | error | kv-transfer, hf3fs, import, dependencies, vllm |
| Hf3fsClient.check Failed | exception | error | kv-transfer, hf3fs, validation, bounds-check, vllm |
| Rank {rank} not initialized | exception | error | kv-transfer, hf3fs, metadata-server, initialization, vllm |
| Invalid initialization parameters | http | error | kv-transfer, hf3fs, metadata-server, http-400, validation, vllm |
| Invalid request format: need 'rank' and 'keys' | http | error | kv-transfer, hf3fs, metadata-server, http-400, validation, vllm |
| Allocation failed: {str(e)} | http | error | kv-transfer, hf3fs, metadata-server, http-500, vllm |
| Invalid request format: need 'rank' and 'confirmations' | http | error | hf3fs, metadata-server, http-400, request-validation, kv-transfer |
| Confirmation failed: {str(e)} | http | error | hf3fs, metadata-server, http-500, state-corruption, kv-transfer |
| Invalid keys format | http | error | hf3fs, metadata-server, http-400, request-validation |
| Key existence check failed: {str(e)} | http | error | hf3fs, metadata-server, http-500, kv-transfer |
| Failed to get key locations: {str(e)} | http | error | hf3fs, metadata-server, http-500, kv-transfer |
| Failed to connect to metadata server: {e} | exception | critical | hf3fs, metadata-server, network, connection-failure, kv-transfer |
| Attempted to free more buffers than allocated | exception | error | hf3fs, buffer-pool, double-free, kv-transfer, resource-management |
| Either vllm_config must be provided, or all of model_config, | validation | error | lmcache, config, api-misuse, kv-transfer |
| Unsupported new_block_ids type {type(new_block_ids)}: should | exception | error | lmcache, version-mismatch, type-validation, scheduler, kv-transfer |
| MLA only works with naive serde mode.. | validation | error | lmcache, mla, config, serde, kv-transfer |
| layerwise MLA connector is not supported yet | validation | error | lmcache, mla, layerwise, config, kv-transfer |
| Request {req_id} is not in _unfinished_requests, but it is s | exception | error | lmcache, scheduler, state-inconsistency, kv-transfer |
| LMCacheMPConnector only works without hybrid kv cache manage | exception | error | lmcache, hybrid-kv-cache, config, kv-transfer, cli-flag |
| Unknown KVConnectorRole: {self.role} | exception | error | lmcache, enum, version-mismatch, config, kv-transfer |
| Mooncake is not available | exception | critical | mooncake, import-error, dependencies, rdma, kv-transfer |
| Mooncake Transfer Engine initialization failed. | exception | critical | mooncake, rdma, initialization, config, kv-transfer |
| No KV cache tensors were registered with Mooncake. | exception | critical | mooncake, kv-cache, registration, kv-transfer |
| Mooncake batch memory registration failed. | exception | critical | mooncake, rdma, memory-registration, gpu, kv-transfer |
| Engine ID mismatch for dp_rank={payload.dp_rank}: expected { | http | error | mooncake, bootstrap-server, engine-id, dp-rank, http-400, kv-transfer |
| Worker with dp_rank={payload.dp_rank}, tp_rank={payload.tp_r | http | error | distributed, kv-transfer, mooncake, registration, http-400 |
| Mooncake preferred_segment override must be a non-empty stri | validation | error | mooncake, rdma, config, validation |
| MooncakeStoreConnector does not support: {unsupported} | validation | error | mooncake, kv-transfer, hybrid-attention, mamba, context-parallel, config |
| block_size ({self.block_size}) must be a multiple of hash_bl | validation | error | mooncake, kv-transfer, block-size, config |
| Group count mismatch: tracker has {len(self.allocated_block_ | exception | error | mooncake, kv-transfer, internal-invariant, scheduler |
| Request {req_id} is not in _unfinished_requests | exception | error | mooncake, kv-transfer, scheduler, race-condition, internal-bug |
| unknown Mooncake mode: {self.mode!r} | validation | error | mooncake, config, validation |
| local_buffer_size must be > 0 | exception | error | mooncake, config, validation |
| embedded mode requires global_segment_size > 0 | exception | error | mooncake, config, rdma, validation |
| standalone-store mode requires global_segment_size == 0 | exception | error | mooncake, config, validation |
| The environment variable 'MOONCAKE_CONFIG_PATH' is not set. | exception | error | mooncake, environment, config, startup |
| tenant_id must be a string or null, got {type(value).__name_ | exception | error | mooncake, config, type-error |
| Unsupported type for size: {type(value)} | exception | error | mooncake, config, type-error, parsing |
| Size cannot be empty. | exception | error | mooncake, config, parsing |
| Invalid format: '{value}' | exception | error | mooncake, config, parsing |
| Invalid numeric value '{number_str}' in: '{value}' | exception | error | mooncake, config, parsing |
| Partial-tail offloads for one request must share a boundary | exception | error | mooncake, kv-transfer, internal-invariant, hybrid-attention |
| Please install mooncake by following the instructions at htt | exception | critical | mooncake, import-error, installation, kv-transfer |
| Initialize MooncakeDistributedStore failed. | exception | critical | mooncake, rdma, initialization, network, native-error |
| Invalid MoRIIO backend {backend!r} in kv_connector_extra_con | exception | error | moriio, kv-transfer, config, rdma, validation |
| Malformed zmq_address {zmq_address!r}: expected 'host:IP,han | exception | error | zmq, kv-transfer, config, validation |
| request_id {request_id!r} does not embed a peer zmq_address | exception | error | kv-transfer, config, routing, disaggregated-prefill |
| request_id {request_id!r} does not embed a peer zmq_address | exception | error | kv-transfer, config, validation, disaggregated-prefill |
| Unexpected socket type: {socket_type} | exception | error | zmq, internal-api, validation |
| TP sizes must be positive | exception | error | tensor-parallel, kv-transfer, config, validation |
| local_tp_rank {local_tp_rank} must be in [0, {local_tp_size} | exception | error | tensor-parallel, kv-transfer, validation |
| remote tp_size {remote_tp_size} must be a multiple of local | exception | error | tensor-parallel, kv-transfer, heterogeneous-tp, config |
| local tp_size {local_tp_size} must be a multiple of remote t | exception | error | tensor-parallel, kv-transfer, heterogeneous-tp, config |
| TP sizes and total_num_kv_heads must be positive | exception | error | tensor-parallel, kv-transfer, model-config, validation |
| MoRIIO heterogeneous TP requires replicated KV heads on both | exception | error | tensor-parallel, kv-transfer, gqa, heterogeneous-tp, model-config |
| consumer tp_size {consumer_tp_size} must be a multiple of pr | exception | error | tensor-parallel, kv-transfer, ack, heterogeneous-tp, config |
| request {request.request_id!r}: request_id has no embedded p | exception | error | kv-transfer, config, routing, worker, disaggregated-prefill |
| MoRIIO is not available. Please ensure the 'mori' package is | exception | critical | dependencies, mori, kv-transfer, environment |
| Ping failed after {retry_count} retries | exception | critical | network, zmq, kv-transfer, connectivity, retry |
| handshake failed, unexpected msg type | exception | error | zmq, handshake, protocol-mismatch, kv-transfer |
| Unexpected frame! {received_frame = } | exception | error | zmq, handshake, protocol-mismatch, kv-transfer |
| unexpected frame! {received_frame = } | exception | error | zmq, handshake, protocol-mismatch, kv-transfer |
| MoRIIO KV cache block size mismatch for layer {layer_name}: | exception | error | kv-cache, model-config, hybrid-attention, kv-transfer |
| Eager MoRIIO handshake failed for {remote_engine_id} on at l | exception | critical | handshake, tensor-parallel, network, kv-transfer, distributed |
| Input list lengths mismatch | exception | error | |
| Failed to register local memory: {e} | exception | error | |
| {len(errors)}/{len(transfers_to_wait)} transfers failed: | exception | error | |
| Error processing message: {e} | exception | error |