🐛 Regression: Jenkins EC2 spike agent template not provisioned when base is exhausted (since 2049)

[zendesk-abhijeet]

Jenkins and plugins versions report

Environment

Plugin version(s):

• Works: 2045.v06da_da_a_46422
• Broken: 2049.v127f0f109c29

Jenkins version: 2.541.2
Cloud: EC2

What Operating System are you using (both controller, and any agents involved in the problem)?

Linux Ubuntu for both. Jenkins controller is deployed in a k8s pod.

Reproduction steps

1. Configure EC2 cloud with two templates that share the same label, e.g. my_agent.
   • Template 1: description my_agent base, instance cap small (e.g. 1)
   • Template 2: description my_agent spike, instance cap > 0
2. Run jobs that require label my_agent so Template 1 reaches its cap.
3. Observe that in 2045, Template 2 provisions; in 2049, Template 2 does not provision and jobs remain queued.

Expected Results

When multiple templates match a label, provisioning should try them in order:

1. Attempt provisioning using the first template
2. If it cannot satisfy demand (e.g., no possible new slaves / cap reached), continue to the next matching template. This is the behaviour we observed in 2045.

Actual Results

Only the first matching template appears to be considered for provisioning; the second template is not attempted, so spike capacity is not used even when base is exhausted.

Anything else?

• Both templates are eligible for the label my_agent.
• Base capacity is fully utilised at time of failure.

Are you interested in contributing a fix?

Can contribute with a patch to fix the issue.