SignalR Auto Scale In does not trigger when condition meet

Question

SignalR Auto Scale In does not trigger when condition meet

pipi 25

I have setup a metric conditions for scale in/out of SignalR. The scale out works perfectly but it will not scale in. The conditions was at below:

User's image

When the ConnectionCount (Max) is less than 930, it does not trigger. May I know how to resolve the problem?

Golla Venkata Pavani 3,810 Reputation points Microsoft External Staff Moderator

2026-03-02T12:39:37.1933333+00:00

Hi @pipi

Thank you for reaching us regarding you have got your SignalR autoscale rule set up to scale in when the connection count is ≤ 930 for 5 minutes, but you’re not seeing the scale-in kick off.

Based on the screenshots you provided, the scale‑in condition is being evaluated correctly, but Azure Autoscale cannot reduce the SignalR unit count because the minimum capacity is set to 2, which is also the current capacity. According to Azure Monitor autoscale behavior, autoscale will never scale below the configured minimum value, even if the metric threshold is met. To allow scale‑in, the minimum unit count must be set lower than the current unit count.

Kindly let us know If you have any additional details or concerns in the meantime.

Please "upvote" if the information helped you. This will help us and others in the community as well.
pipi 25 Reputation points

2026-03-03T00:34:51.5333333+00:00

hi Pavani,

Thank you for the reply, but i think you overlook the minimum instance, it is 1 not 2.
Golla Venkata Pavani 3,810 Reputation points Microsoft External Staff Moderator

2026-03-04T18:26:11.76+00:00

Hi @pipi

I'm just reaching out to see if your issue has been resolved or if you've had a chance to review my previous comment?

2 answers

Your answer

Golla Venkata Pavani 3,810 Reputation points Microsoft External Staff Moderator

2026-03-02T12:39:37.1933333+00:00

Hi @pipi

Thank you for reaching us regarding you have got your SignalR autoscale rule set up to scale in when the connection count is ≤ 930 for 5 minutes, but you’re not seeing the scale-in kick off.

Based on the screenshots you provided, the scale‑in condition is being evaluated correctly, but Azure Autoscale cannot reduce the SignalR unit count because the minimum capacity is set to 2, which is also the current capacity. According to Azure Monitor autoscale behavior, autoscale will never scale below the configured minimum value, even if the metric threshold is met. To allow scale‑in, the minimum unit count must be set lower than the current unit count.

Kindly let us know If you have any additional details or concerns in the meantime.

Please "upvote" if the information helped you. This will help us and others in the community as well.
pipi 25 Reputation points

2026-03-03T00:34:51.5333333+00:00

hi Pavani,

Thank you for the reply, but i think you overlook the minimum instance, it is 1 not 2.
Golla Venkata Pavani 3,810 Reputation points Microsoft External Staff Moderator

2026-03-04T18:26:11.76+00:00

Hi @pipi

I'm just reaching out to see if your issue has been resolved or if you've had a chance to review my previous comment?

Answer 1

Hi @pipi

Thank you for the clarification and apologies again for any confusion.

The likely cause is a mismatch in your scale-in rule: Time grain statistic = Maximum but Time aggregation = Average. This evaluates the average of per-minute maximums over 5 minutes, so even minor spikes can keep the value >930, blocking scale-in despite low averages.

Quick fix:

Change Time grain statistic to Average (recommended for consistent low-load detection).
Optionally set Time aggregation to Average too, and increase Duration to 10 minutes.

Microsoft recommends using Connection Quota Utilization (%) for rules instead of raw Connection Count (e.g., scale in <20–30%) for better reliability and headroom.

Save, then monitor Autoscale history under low load. If it still doesn't trigger, check Metrics (Max aggregation, 1-min grain) and recent scale events.

Kindly let us know If you have any additional details or concerns in the meantime.

Please "upvote" if the information helped you. This will help us and others in the community as well.

Answer 2

Hello , Welcome to MS Q&A

The most likely issue is a time aggregation mismatch in the autoscale configuration.

Currently, the scale-in rule is configured as:

Time Grain Statistic: Maximum

Time Aggregation: Average

This mismatch prevents the scale-in action from triggering correctly.

Why This Is a Problem

You are checking whether:

Connection Count (Maximum) ≤ 930

However, Azure evaluates the average of the maximum values over the selected duration.

This means that if even one minute within the evaluation period has a maximum value greater than 930, the calculated average may still exceed 930 — even if the current maximum value is below 930.

As a result, the scale-in condition does not get satisfied.

Quick Fix

Update the scale-in rule so that the aggregation matches the statistic.

Current (Incorrect):

Time Grain Statistic: Maximum

Time Aggregation: Average

Correct Configuration:

Time Grain Statistic: Maximum

Time Aggregation: Maximum

The aggregation must align with the selected statistic.

Recommended Scale-In Configuration

Update the rule as follows:

Metric: Connection Count

Dimension: Endpoint of connections = All values

Operator: Less than or equal to

Threshold: 930

Duration: 10 minutes (increase from 5 minutes)

Time Grain: 1 minute

Time Grain Statistic: Maximum

Time Aggregation: Maximum

Action: Decrease count by 1

Cool Down: 10 minutes (increase from 5 minutes)

Pls check and let us know.

Thanks

Deepanshu

pipi 25 Reputation points

2026-03-02T09:32:11.9433333+00:00

hi Deepanshu,

Thank you for the reply. However, I have tried the solutions you provided but still not working:

Share via

SignalR Auto Scale In does not trigger when condition meet

2 answers

Why This Is a Problem

Quick Fix

Recommended Scale-In Configuration

Your answer