[Standup] Serve models from read-only pvcs #553
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
* The `pvc` holding the models mounted by the `pods` are now **read-only**, with the environment variable `VLLM_CACHE_ROOT` (default value `/tmp/vllm` being automatically set) * Make variables `VLLM_CACHE_ROOT`, `VLLM_WORKER_MULTIPROC_METHOD`, `VLLM_ALLOW_LONG_MAX_MODEL_LEN`, `VLLM_SERVER_DEV_MODE`, `VLLM_LOAD_FORMAT`, `VLLM_LOGGING_LEVEL`, `VLLM_ENABLE_SLEEP_MODE` part of `LLMDBENCH_VLLM_COMMON` and automatically added to the environment (ready to be used by `vllm`) * **IMPORTANT** Any environment variable with the string `LLMDBENCH_VLLM_COMMON_VLLM_<SOMETHING>`, `LLMDBENCH_VLLM_STANDALONE_VLLM_<SOMETHING>`, `LLMDBENCH_VLLM_MODELSERVICE_PREFILL_VLLM_<SOMETHING>`, `LLMDBENCH_VLLM_MODELSERVICE_DECODE_VLLM_<SOMETHING>` will be exported within the `pod` as `VLLM_<SOMETHING>` * Annotations `stood-up-by`, `stood-up-from` and `stood-up-via` are automatically added to both `standalone` and `modelservice` pods * Added a selector for the `python` implementation of `kubectl_get` * Added a `python` implementation of `kubectl_delete` * Extra command line parameters required when `VLLM_ENABLE_SLEEP_MODE` is set to `true` are now rendered inside `add_command_line_options` * Fix for `get_model_name_from_pod`. Get kubeconfig from current context. * More informative error messages when `run.sh` cannot locate a stack * Removed the environment variable `LLMDBENCH_HARNESS_CONTAINER_IMAGE` Signed-off-by: maugustosilva <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
pvcholding the models mounted by thepodsare now read-only, with the environment variableVLLM_CACHE_ROOT(default value/tmp/vllmbeing automatically set)VLLM_CACHE_ROOT,VLLM_WORKER_MULTIPROC_METHOD,VLLM_ALLOW_LONG_MAX_MODEL_LEN,VLLM_SERVER_DEV_MODE,VLLM_LOAD_FORMAT,VLLM_LOGGING_LEVEL,VLLM_ENABLE_SLEEP_MODEpart ofLLMDBENCH_VLLM_COMMONand automatically added to the environment (ready to be used byvllm)LLMDBENCH_VLLM_COMMON_VLLM_<SOMETHING>,LLMDBENCH_VLLM_STANDALONE_VLLM_<SOMETHING>,LLMDBENCH_VLLM_MODELSERVICE_PREFILL_VLLM_<SOMETHING>,LLMDBENCH_VLLM_MODELSERVICE_DECODE_VLLM_<SOMETHING>will be exported within thepodasVLLM_<SOMETHING>stood-up-by,stood-up-fromandstood-up-viaare automatically added to bothstandaloneandmodelservicepodspythonimplementation ofkubectl_getpythonimplementation ofkubectl_deleteVLLM_ENABLE_SLEEP_MODEis set totrueare now rendered insideadd_command_line_optionsget_model_name_from_pod. Get kubeconfig from current context.run.shcannot locate a stackLLMDBENCH_HARNESS_CONTAINER_IMAGE