From 480f3c48c13f95441293ff32294ab10da10c79c3 Mon Sep 17 00:00:00 2001 From: Justin Date: Thu, 15 Oct 2026 10:15:09 -0400 Subject: [PATCH 1/2] docs(flash): explain standby workers next to the workers parameter --- flash/configuration/parameters.mdx | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/flash/configuration/parameters.mdx b/flash/configuration/parameters.mdx index db0005a80..374034265 100644 --- a/flash/configuration/parameters.mdx +++ b/flash/configuration/parameters.mdx @@ -148,6 +148,10 @@ Controls worker scaling. Accepts either a single integer (max workers with min=0 - `workers=(1, N)`: Avoid cold starts by keeping at least one worker warm - `workers=(N, N)`: Fixed worker count for consistent performance + +Runpod also keeps a few standby workers parked for each endpoint so jobs can start quickly with [FlashBoot](/serverless/endpoints/endpoint-configurations#flashboot). The API shows their target count as `workersStandby`, and the [`/health` operation](/serverless/endpoints/operation-reference#health) can list them as ready. Runpod picks this number for you, and you don't pay compute for a parked worker until it starts running. `workers=(0, N)` still scales to zero. + + ### idle_timeout **Type**: `int` From 58240ccb61f4440044f465b25888e233fe965235 Mon Sep 17 00:00:00 2001 From: lgunreddi Date: Fri, 16 Oct 2026 09:04:23 -0400 Subject: [PATCH 2/2] Update parameters.mdx --- flash/configuration/parameters.mdx | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/flash/configuration/parameters.mdx b/flash/configuration/parameters.mdx index 374034265..84ef3abcd 100644 --- a/flash/configuration/parameters.mdx +++ b/flash/configuration/parameters.mdx @@ -149,7 +149,8 @@ Controls worker scaling. Accepts either a single integer (max workers with min=0 - `workers=(N, N)`: Fixed worker count for consistent performance -Runpod also keeps a few standby workers parked for each endpoint so jobs can start quickly with [FlashBoot](/serverless/endpoints/endpoint-configurations#flashboot). The API shows their target count as `workersStandby`, and the [`/health` operation](/serverless/endpoints/operation-reference#health) can list them as ready. Runpod picks this number for you, and you don't pay compute for a parked worker until it starts running. `workers=(0, N)` still scales to zero. +Runpod also keeps a few standby workers parked for each endpoint so jobs can start quickly with [FlashBoot](/serverless/endpoints/endpoint-configurations#flashboot). The API shows their target count as `workersStandby`, and the [`/health` operation](/serverless/endpoints/operation-reference#health) can list them as ready. Runpod picks this number for you, and you aren't billed for a parked worker until it starts running. +`workers=(0, N)` still scales to zero. ### idle_timeout