Azure Data Explorer cluster query endpoint unreachable — queries time out after ~30 seconds ("Network Error… connect to the VPN client")

Darnell Brown 0 Reputation points
2026-11-19T05:36:50.4566667+00:00

I am on the Basic support plan, so I cannot open a technical support case — posting here as the free route.

Cluster: doughboyanalytics-adx, East US

Resource group: DoughBoyAnalytics-RG

Database: DoughBoyAnalytics (tables BikeEvents, BikeRides, BikeClips — all exist, 0 rows)

Cluster URI: https://doughboyanalytics-adx.eastus.kusto.windows.net

Symptom: In the ADX query pane (Azure Portal and the standalone ADX web UI), any query — including .show tables — fails after about 30 seconds with:

Network Error: Please check your internet connection. You may need to connect to the VPN client.

When it happened: four separate attempts — Nov 9, Nov 15, Nov 17, and Nov 19, 2026. Same error every time.

What works:

  • Azure Portal sign-in and control plane work fine; the cluster shows Running.
  • The three tables are visible in the Connections tree.
  • Public network access is enabled from all networks — no firewall/IP rules, no private endpoints.
  • DNS for doughboyanalytics-adx.eastus.kusto.windows.net resolves (20.168.201.207), but TCP 443 never completes. management.azure.com and the ingest host answer fine.
  • It also fails from my own home network and browser: "error loading databases, please try again in a few minutes."

Latest request IDs (Nov 19):

  • Kusto.Web.KWE.Query;d82fa8b3-741b-4fad-a2fd-a73e49ec7361;ff741cc2-777e-49b2-8ac0-ff497b9dbcd8
  • Kusto.Web.KWE.Query;798f77fb-a276-4326-807c-0b502fedf234;a773bc2d-f0b5-4646-bd10-b822aae31d91

What I tried: the portal's "Diagnose and solve problems" troubleshooter for Azure Data Explorer Tools only returned a self-help article about authorization errors and generic browser steps — nothing matching a 30-second timeout with TCP 443 never completing.

Impact: I cannot run queries or ingest my catalog data (3 events, 14 rides, 420 clips), so the project is blocked at the pre-check step. Is the query endpoint for this cluster reachable, and is there anything on the service side blocking the data-plane connection?

Azure Data Explorer
Azure Data Explorer

An Azure data analytics service for real-time analysis on large volumes of data streaming from sources including applications, websites, and internet of things devices.


1 answer

Sort by: Oldest
  1. AI answer

    2026-11-19T05:38:06.12+00:00

    Check the cluster in Azure Data Explorer Resource Health now; if it shows Unavailable, open the support request offered there because platform-unavailability requests do not require a paid technical support plan.^1^

    Determine whether Azure has identified a service-side problem

    1. In the Azure portal, open Monitor → Service Health → Resource health.
    2. Filter:
      • Subscription: the cluster’s subscription
      • Resource type: Azure Data Explorer
    3. Select the affected cluster.
    4. Review both its current status and Health history.

    ADX Resource Health evaluates signals including cluster query availability and query failures. It updates every one to two minutes, and its history covers up to four weeks.^2^

    Interpret the result as follows:

    • Unavailable: Azure has detected an ongoing condition preventing queries or ingestion. Select the support option from this pane.
    • Degraded: Record the recommendation and event details, then escalate if the endpoint remains unusable.
    • Unknown: Resource Health has not received cluster information for more than ten minutes. Since this cluster is reportedly running, persistent Unknown is significant and should be reported.
    • Available: Select Report incorrect health status, because the observed data-plane endpoint is still inaccessible.^1^

    Also open Monitor → Service Health → Service issues, filter for Azure Data Explorer and East US, and inspect active and resolved events. This view identifies affected regions, subscriptions, and individual resources; resolved events remain in Health history for 90 days.^3^

    The public Azure status page alone cannot rule out this problem because it reports widespread incidents, whereas Service Health and Resource Health expose subscription- or resource-specific impact.^4^

    Check ADX platform telemetry without querying the endpoint

    Because the query endpoint cannot complete TCP 443, .show diagnostics cannot currently be used. Instead:

    1. Open the cluster resource.
    2. Select Monitoring → Metrics.
    3. Set the time range to include each failure window.
    4. Add Keep Alive and inspect whether it remains close to 1 or drops around the failures. Microsoft documents a value close to 1 as healthy.^5^
    5. Add relevant query, ingestion, CPU, and resource metrics where available.

    If Keep Alive drops while TCP 443 is timing out, include the chart in the escalation. If it remains healthy, report that discrepancy as well; it helps isolate a front-end endpoint or routing failure from general cluster compute health.

    Diagnostic logs can capture queries and commands that reach a final state, but they may contain no matching entry when the network connection never reaches the service. They are therefore supporting evidence, not a prerequisite for this case.^6^

    What the existing evidence rules out

    The supplied URI has the documented ADX format, the cluster is running, and Enabled from all networks permits access through the public endpoint.^7^

    Because:

    • DNS resolves,
    • TCP 443 to the query host does not complete,
    • the same failure occurs from separate networks and both ADX interfaces,
    • even .show tables fails, and
    • no HTTP authorization response is returned,

    this is not behaving like a Kusto query-duration or database-permission failure. Increasing Settings → Connection → Query timeout would not repair a TCP connection that never completes; that setting controls query and administrative-command time limits after connectivity exists.^8^

    The browser’s VPN wording is generic and does not establish that a VPN is required. Do not apply VPN Gateway, VM NSG, or private-endpoint remediation to this cluster unless its configuration changes to use those components.

    Escalate with the Basic plan

    If Resource Health reports Unavailable, create the request directly from that pane. Microsoft states that no support agreement is required when a resource is unavailable because of a platform event.^1^

    If the portal permits manual creation, use:

    Cluster → Support + Troubleshooting → Create a support request

    Select the closest ADX category to cluster/database access or unable to connect, then include:

    • cluster resource ID and East US region;
    • UTC timestamps for all four occurrences;
    • the two request IDs;
    • Resource Health status/history and any Service Health tracking ID;
    • Keep Alive screenshots or exported metric data;
    • confirmation that public access is enabled from all networks;
    • confirmation that the query hostname resolves but TCP 443 times out from independent networks;
    • confirmation that the ingest and ARM endpoints remain reachable;
    • business impact: query and ingestion paths blocked.

    Creating a request requires Owner, Contributor, Support Request Contributor, or equivalent Microsoft.Support/* permission at subscription scope.^9^

    There is not enough public information to verify the live reachability of this individual cluster or identify a service-side block. Resource Health and Microsoft’s backend telemetry tied to the request IDs are the authoritative next checks.


    References

    1. Azure Resource Health FAQ
    2. Monitor Azure Data Explorer using Resource Health (preview)
    3. View Service issues - Azure Service Health | Microsoft Learn
    4. What to do during an Azure service disruption
    5. Check the health of an Azure Data Explorer cluster
    6. Monitor Azure Data Explorer
    7. Manage public access to your Azure Data Explorer cluster
    8. Customize settings in the Azure Data Explorer web UI
    9. Create an Azure support request
    AI-generated content may be incorrect. Read our transparency notes for more information.

    Was this answer helpful?

Your answer

Answers can be marked as 'Accepted' by the question author and 'Recommended' by moderators, which helps users know the answer solved the author's problem.