> For the complete documentation index, see [llms.txt](https://docs.vega365.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.vega365.com/tandem/recovering-from-recorder-failure/smart-client.md).

# Smart Client

What an operator sees when the attached recorder fails, and what they have to do about it.

## The operator experience during failover

When the attached recording server fails, the headless member takes over, and clients resume from it. This page shows what that looks like from Smart Client.

The operator does nothing. Three things happen on their own:

1. **Live video fails over.** Views automatically resume from the other member, with the operator's workspace unchanged.
2. **The full playback history is available.** The member that took over holds every camera's complete history.
3. **No video is lost at the moment of failure.** Recording runs straight through the instant the recorder fails.

{% embed url="<https://youtu.be/dE0jtJB5FdU>" %}
Smart Client during a failover of the attached recorder.
{% endembed %}

### 1. Live video fails over automatically

#### Before the failure

When the attached member fails, live video pauses briefly while Smart Client reconnects, then resumes from the other member. The operator takes no action.

<figure><img src="https://92057524-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2Fi0mCLda4cGtI3Jwb6IQA%2Fuploads%2F6O6xoMocwKmjt18hadGp%2Fsc_01_live_before.png?alt=media&amp;token=db9f9870-5091-49a5-b86b-8f750a68b14c" alt=""><figcaption><p>Live video before the failure</p></figcaption></figure>

#### The recorder fails

The tiles stop showing video and report that the connection to the server is lost and the client is reconnecting. Each tile names its camera and the recording server address it was connected to.

<figure><img src="https://92057524-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2Fi0mCLda4cGtI3Jwb6IQA%2Fuploads%2FaG2pDhoUDZt1HM6OVlgb%2Fsc_02_connection_lost.png?alt=media&amp;token=7ef1247d-590d-4def-b335-583532e7a32e" alt=""><figcaption><p>Connection lost</p></figcaption></figure>

#### Live video returns

All four tiles return to live video. The view, layout, camera names, and camera tree are exactly as they were—the operator's workspace is untouched.

<figure><img src="https://92057524-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2Fi0mCLda4cGtI3Jwb6IQA%2Fuploads%2FxehpeKIJzjhNTfYNAayT%2Fsc_04_live_after.png?alt=media&amp;token=63b6dd42-f9b1-4fb3-88c0-7cbf94cde2f3" alt=""><figcaption><p>Live video restored, from the other member</p></figcaption></figure>

### How long it takes

It takes a few seconds after the primary goes offline for the secondary to start serving clients.  If you need near instantaneous live video recovery instead, consider our other product, [Redundancy Management Framework](https://docs.vega365.com/redundancy-management-framework/).

### 2. The full playback history is available after failover

After failover, the member serving clients holds the complete recording history—every camera, going back as far as retention allows. Not only what was recorded since it took over.

It was recording every camera long before the failure, so its media database already contains the history. An operator can switch to playback and scrub back through the hours and days before the failure exactly as before. Search, export, and evidence lock work against that same history.

<figure><img src="https://92057524-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2Fi0mCLda4cGtI3Jwb6IQA%2Fuploads%2Fq0OXYtKvslgDUJHFxFL2%2FPlayback_timeline_primary.png?alt=media&amp;token=9ef0abd5-f7a0-495d-ba16-70580f8ccb32" alt=""><figcaption></figcaption></figure>

<figure><img src="https://92057524-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2Fi0mCLda4cGtI3Jwb6IQA%2Fuploads%2F89HX8zDAGnlZ8Sf99VWz%2FPlayback_timeline_secondary.png?alt=media&amp;token=8eaa1a54-b559-44b8-aa82-e64a993390c9" alt=""><figcaption><p>After the failover — the same history, now served by the other member</p></figcaption></figure>

The two images show the same period of history. Before the failover, it comes from the original attached member; after, from the member that took over. The content is the same.

{% hint style="info" %}
This assumes the two members were in sync before the failure. If the member now serving was down earlier, and its gap was never backfilled—because backfill was disabled or media encryption is in use—that earlier gap is absent from its history.
{% endhint %}

### 3. No video is lost at the moment of failure

Tandem is **active-active**. Both members of a cluster record every camera, continuously, at the same time. When the attached member fails, the other does not have to start recording—it has been recording all along, and it keeps recording straight through the failure.

The picture below shows the primary recorder's playback timeline, with the timestamp at the moment of failure. Backfill is off, so there is a recording gap until the recorder is restored later.

<figure><img src="https://92057524-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2Fi0mCLda4cGtI3Jwb6IQA%2Fuploads%2FOmNsmkpzDHBh6UevXAiX%2FP_failure_time_stamp.png?alt=media&amp;token=5049c351-bcc1-4ded-bf45-f645df3423bb" alt=""><figcaption><p>Primary timeline at the moment of failure</p></figcaption></figure>

The picture below shows the secondary recorder's playback timeline, with the timestamp at the moment the primary failed. Because it is active-active, there is no recording gap, and content is not lost.

<figure><img src="https://92057524-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2Fi0mCLda4cGtI3Jwb6IQA%2Fuploads%2F342hWZ3z4ttayEmE78Rm%2FS_no_gaps.png?alt=media&amp;token=ead9ab25-837a-47e0-aa41-527a14da2948" alt=""><figcaption><p>Secondary recorder timeline with the timestamp at the moment of failure of primary shows no gaps.</p></figcaption></figure>

### What the administrator sees

On the Clustering **Dashboard**, the cluster moves to **Failed over**, and the roles swap: the former attached member shows **Standby**, the other shows **Active**. Both clustering agents continue running.

<figure><img src="https://92057524-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2Fi0mCLda4cGtI3Jwb6IQA%2Fuploads%2FjyL1RDiCEpi9LyoXFeDL%2FFailover_dashboard.png?alt=media&amp;token=206272d7-a57f-453a-918b-4dbd9daeed53" alt=""><figcaption><p>Clustering Dashboard — failed over</p></figcaption></figure>

The configured Primary and Secondary roles do not change. What changes is which member is attached and serving clients.

### What does not happen

* **No operator action.** No reconnecting, no reselecting cameras, no switching views.
* **No changes to views or layouts.** Saved views continue to work untouched.
* **No gap in the recordings.** The member taking over was already recording every camera.&#x20;
* **No loss of playback history.** Footage from before the failure remains available for playback, search, and export.
* **No reconfiguration of anything connected to the VMS.** The cluster keeps one recorder identity across the failover.
