xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2026-05-22 08:31:18 +08:00
Code Issues Packages Projects Releases Wiki Activity

[BugFix] Unblock use of LoRA with data parallel mode (#30220)

Browse Source 
Signed-off-by: Nick Hill <nhill@redhat.com>
This commit is contained in:
Nick Hill 2025-12-07 20:21:05 -08:00 committed by GitHub
parent 344b50d525
commit d726a7b0ed
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 4 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
vllm/v1/metrics/loggers.py
View File

@ -952,7 +952,10 @@ class PrometheusStatLogger(AggregateStatLoggerBase):
self.gauge_lora_info: Gauge | None = None self.gauge_lora_info: Gauge | None = None
if vllm_config.lora_config is not None: if vllm_config.lora_config is not None:
if len(self.engine_indexes) > 1: if len(self.engine_indexes) > 1:
raise NotImplementedError("LoRA in DP mode is not supported yet.") logger.warning(
"vllm:lora_requests_info prometheus metrics may be "
"incorrect/misleading with data parallel deployments."
)
self.labelname_max_lora = "max_lora" self.labelname_max_lora = "max_lora"
self.labelname_waiting_lora_adapters = "waiting_lora_adapters" self.labelname_waiting_lora_adapters = "waiting_lora_adapters"
self.labelname_running_lora_adapters = "running_lora_adapters" self.labelname_running_lora_adapters = "running_lora_adapters"