fix: add validation to reject leading/trailing whitespace in target field

[VikaCep]

Summary

• Adds validation to reject leading/trailing whitespace in the target/instance field
• Prevents a bug where trailing whitespace causes dashboard metrics to fail to load (see https://github.com/grafana/support-escalations/issues/20484)

Problem

A customer reported that after duplicating a browser check, the dashboard showed no metrics even though the check was running successfully. Investigation revealed:

1. The check was saved with a trailing space in the Instance field: "FRANCE " instead of "FRANCE"
2. Metrics were generated with instance="FRANCE " (with trailing space)
3. When viewing the dashboard, URL parameter handling strips trailing whitespace: var-instance=FRANCE
4. The probe variable query sm_check_info{instance="$instance"} uses "FRANCE" (no space)
5. No match -> empty probe dropdown -> dashboard shows "N/A" for all metrics

The main Checks list page displayed data correctly because it queries metrics directly without the URL variable interpolation.

Solution

Add validation to show an error when the target field contains leading or trailing whitespace:

"Target cannot have leading or trailing whitespace"

This approach:

• Forces users to manually fix the whitespace, making them aware of the change
• Avoids silently transforming values which could break Terraform configs or metric continuity
• Matches the pattern used by jobSchema for rejecting invalid characters

[g3john]

Awesome! What if we just trimmed the value for them, would that cause unexpected results?

[VikaCep]

Awesome! What if we just trimmed the value for them, would that cause unexpected results?

I thought about removing the trailing whitespaces automatically on save, but in the end I decided to display a validation error message instead, because when you change job/target values (unique identifiers for a check) we lose continuity on metrics, so I didn't want that change to happen without customers noticing when they change a different check value.

For this particular customer I did suggest them to remove the whitespace themselves to fix their issue.

[github-actions]

Script size changes

Name	+/-	Main	This PR	Outcome
[263.js]	=	1,715.58 kB	1,715.54 kB	✅
[854.js]	New file	-	799.86 kB	✅
[datasource/module.js]	=	25.00 kB	25.00 kB	✅
[692.js]	=	20.56 kB	20.56 kB	✅
[663.js]	=	5.82 kB	5.82 kB	✅
[module.js]	=	4.90 kB	4.90 kB	✅
[234.js]	Deleted file	805.52 kB	-

Totals

Name	+/-	Main	This PR	Outcome
[Scripts]	-0.22%	2,577.39 kB	2,571.69 kB	✅
[Non-script Assets]	=	2,675.38 kB	2,675.38 kB	✅
[All]	-0.11%	5,252.76 kB	5,247.07 kB	✅

Generated by 🚫 dangerJS against 4ff7d7f

[g3john]

LGTM!

[ckbedwell]

I'm fine with this but this is a scenes bug as far as I'm concerned (all the more reason to drop it 😬)

You could end up in a situation where a user was managing a check via the UI and Terraform/API and you get two different validations.

It's pretty edge-casey but as this arose from an escalation originally it's not out of the realm of possibility.

[VikaCep]

Even if this is a scenes bug when forming the dashboard URL, I think trailing whitespaces shouldn’t be allowed in job/target fields as it causes confusing scenarios that are hard to spot.
I think the API/Terraform should enforce this same behavior, which means also adding this validation on their side.