Repository navigation
Conversation
…onnector MT.1098 tells the operator to complete setup in Intune and with the partner whenever a connector reports `notSetUp`. For the Windows Security Center connector with **Connect Windows devices for MAM** already on, that setup is done: Intune keeps the state at `notSetUp`, and sends no heartbeat, until the first Windows MAM user enrols (maester365#2595, measured on a lab tenant). The verdict does not move. Microsoft documents `notSetUp` as blocking device threat messages whatever the cause, so the connector is still not carrying the signal it exists to carry. What changes is the Reason column and the heartbeat line, so the row names what is actually missing. Both guards are anchored: the connector's own ID and its own toggle. Either one alone keeps the generic explanation, which the two new negative cases assert.
|
No actionable comments were generated in the recent review. Walkthrough
Merge Risk:
|
Up to standards ✅🟢 Issues
|
|
Ran MT.1098 from this branch (690ecda) against the tenant from #2595, November 11, 15:30 UTC:
Result: Failed. On live data, the Windows Security Center connector gets the new reason, and the two other notSetUp connectors keep the generic one. It's the same Windows Security Center connector as in #2595, still notSetUp with no heartbeat, so the window with the toggle on is now nearly two days. It went on in the evening of November 9; the reading I posted in #2595 covered about 24 hours. The ubuntu build-validation failure isn't from this PR. The File encoding failure is Two small things, your call:
Thanks for the credit. |
|
Thank you @julien-ly — that is the measurement I could not take, and it lands exactly where the guard aims. Two things your table shows beyond "the text changed":
So the row now names the one action that resolves it: enrol a Windows MAM user, or remove the connector if Windows MAM is not in use. |
|
Measured the transition on the same lab tenant today, with the tenant's first Windows MAM enrolment (main Before (MT.1098 at 20:31 UTC): Windows Security Center Enrolment on an unmanaged Windows 11 device with Edge 157, for a test user targeted by a Windows app protection policy and by a Conditional Access policy that requires an app protection policy or a compliant device. Times in UTC, read from Graph every two minutes:
So the first heartbeat came seven seconds after the first enrolment, well inside the 30 minutes the admin center mentions (one tenant, one enrolment). After (MT.1098 at 21:01 UTC): the Windows Security Center row is identical on main and on this PR: That matches the case the PR describes: |
…d" claim Two corrections to the Windows Security Center case, both about claiming more than the source supports. The guide called it "a documented special case". The live Learn page for the connector says the opposite - Available after creation, Enabled once the toggle goes on - and the only source for the enrolment trigger is the note in the Intune admin center. It now says "a special case noted in the Intune admin center". The Reason opened with "Windows MAM is on". What the check reads is the connector's own windowsMobileApplicationManagementEnabled, the Connect Windows devices for MAM toggle, which says nothing about whether a Windows app protection policy is deployed. The Reason now names that toggle. No behaviour change: the guard, the verdict and the heartbeat line are untouched, and the unit tests match on "first Windows MAM user enrols", which both wordings keep.
|
Thank you @julien-ly — that is the half of the case I could not reach. The branch asserts that Your two points from the earlier review are in
Both are the same mistake on my part — claiming more than the field or the source supports — and neither changes behaviour. The guard, the verdict and the heartbeat line are untouched; the unit tests match on "first Windows MAM user enrols", which both wordings keep. 26 of 26 pass here through On the Linux leg, your reading matches mine exactly, down to the file names. The byte order mark is #2795 and the three |
Closes the remaining part of #2595.
What is wrong today
MT.1098 reports a
notSetUpconnector with: "Setup is incomplete; device threat messages are blocked. Complete setup in Intune and with the partner, or remove an unused connector."For the Windows Security Center connector that advice is wrong in a specific, reproducible case. @julien-ly measured it in a lab tenant (#2595): after turning on Connect Windows devices for MAM,
windowsMobileApplicationManagementEnabledistruewhilepartnerStatestaysnotSetUpandlastHeartbeatDateTimestays0001-01-01T00:00:00Z, unchanged over two days with no Windows device registered for MAM. The Intune admin center says so in the connector blade, quoted by @Mynster9361 in the issue: the status does not show Enabled until the first MAM user enrollment, and updates within 30 minutes of one.So the operator is sent back to the partner to finish a setup that is already finished.
What this does not change
The verdict. Mobile Threat Defense integration with Intune documents
notSetUpas blocking device threat messages since 2309, whatever the reason for the state. A connector in that state is not carrying the signal it exists to carry, and the remediation is real: enroll a MAM user, or remove a connector you are not going to use.FailedstaysFailed.If either of you reads this case as something that should pass, say so and I will change it instead — that is the one part of this where I am taking a position rather than reporting a measurement.
One thing worth flagging
Two Microsoft pages disagree about this connector, which is probably why the issue was confusing to begin with:
notSetUpuntil the first MAM user enrollmentNeither page matches what the API returns. This PR follows the tenant.
The change
notSetUp, only when the connector ID is the Windows Security Center one and its own Windows MAM toggle is on: ask for the enrollment, and say the block is in force until then.Test.MT.1098.mddocuments the case.Verification
Run through
powershell/tests/pester.ps1, so on the Pester the repository pins (5.7.1), not whatever is newest on the machine:The tests are not passing for free: disabling the new condition fails exactly the two cases that assert the new reason, the focused one and the engine one, and nothing else.
Credit where it is due — @Mynster9361 found it and @julien-ly measured it, so I added both to
Contributoralong with @FiraSenax, whose #2736 put the Reason column there in the first place.Summary by CodeRabbit