Monitoring
Monitoring is how you tell whether a connection is doing its job right now and what to do when it is not. The Portal gives you a fleet-level list (the Adapters page, titled Adapter Fleet), a per-connection Operational Console for drilling in, and the operator controls needed to act without leaving the page. Message-level history lives in the Global Event Log.
The three signals to check
When reading an adapter in production, check three signals in order. If one is off, the next signal tells you where to look.
| Signal | What it is | Why it matters | Act when |
|---|---|---|---|
| Operational Status | The live state the platform reports for the running instance: Running, Active, Degraded, Error, Provisioning, Releasing, Standby, Paused, Ports Exhausted, or Inactive. | Tells you whether the connection is up and healthy. Everything else is secondary if the status is not Running. | Operational Status is anything other than Running or Active. |
| Admin State vs Operational Status | Admin State is your intended state (enabled or disabled). Operational Status is what is actually happening. The Portal flags the mismatch for you. | A running connection whose intent and reality disagree is the platform telling you it tried and could not finish, or that something is running that should not be. | Admin State is enabled while Operational Status is Inactive, or Admin State is disabled while the instance is still running. |
| Event log activity | The per-connection narrative in the Global Event Log, newest first: connected, disconnected, messages, warnings, errors. | Operational Status is the current state; the event log is the narrative that explains how you got there. | There is an Error-severity entry, or expected entries (logon, message traffic) have stopped appearing. |
The Portal does not chart message throughput, error rate, or latency. To judge whether traffic is flowing and at what error level, read the event log. The Global Event Log includes a volume-over-time view that buckets loaded entries by severity, which is the fastest way to spot a burst of errors or a sudden drop in activity.
The Adapters list
The Adapters page (titled Adapter Fleet) is the fleet-level view. Read it to find anything that needs attention, then click in to act.
| Column | What it tells you | What to do with it |
|---|---|---|
| Adapter Name | The connection's display name. | Click it to open the Operational Console. |
| Logical ID | The stable identifier used in configuration and bundles. | Search matches on name and logical id. |
| Admin State | Your intended state: enabled or disabled. Differs from Operational Status, which is actual health. | Flag rows where Admin State is enabled but Operational Status is not Running or Active. |
| Operational Status | The live resolved status, using the same labels as the console. While the list is polling it reads Checking…; if the instance cannot be reached it reads Unreachable; during a bulk action it reads Transitioning. | Scan for anything off-normal. |
| Mode | Single Writer or Scale Out. | Combine with status to judge severity. A Single Writer configuration in Error is an outage for that connection. A Scale Out configuration spreads work across several actively claimed instances, so one instance in Error may still leave others carrying traffic. |
| Actions | A per-row menu. | Open the Operational Console, jump to Configuration, or Delete the adapter. |
Reading the Operational Status column
| Status | What it means | Action |
|---|---|---|
| Running | Active and healthy. | No action. |
| Active | Active, with no separate health signal reported. | Treat as healthy unless the event log says otherwise. |
| Degraded | Active but reporting warnings. | Open the console and read the event log for the warnings. |
| Error | The instance reported an unhealthy state, or an attempt to run failed. | Open the console and read the event log entry immediately before the Error; its Description and Details name the cause. Resolve protocol-specific events via the owning plugin's reference. |
| Provisioning / Releasing (transient) | The platform is placing or removing the adapter. | Wait for it to settle. |
| Standby | The instance is awaiting the primary lease. Normal for a redundant Single Writer configuration where another instance is currently active. | No action unless it never becomes Active when it should. |
| Paused | An operator paused the adapter. | Resume it from the console when ready. |
| Ports Exhausted | No available ports in the registry. | Free a port or widen the configured port range. |
| Inactive | The adapter is not running. | If Admin State is enabled, this is a fault: confirm there is adapter-server capacity and read the event log. If Admin State is disabled, this is expected. |
| Unreachable (list only) | The platform cannot reach the adapter instance. | Investigate connectivity between the adapter server and the platform. |
Filter the list by Admin State (All statuses, Enabled, Disabled) or search by name or logical id. There is no filter on Operational Status, so scan the Operational Status column for off-normal rows, or open the Global Event Log filtered to Error severity.
Select rows with the checkboxes to act in batches. The bulk bar offers Start, Stop, Restart, Shutdown, and Delete. Start enables the selection; Stop disables it; Restart issues a shutdown and the platform restarts the instance automatically; Shutdown disables and stops the instance (it will not restart until re-enabled). Restart, Shutdown, and Delete require typing CONFIRM before they run.
For fleet-wide triage, open the Global Event Log and filter to Error severity. It aggregates events across every connection, which is faster than scanning the full list row by row.
The Operational Console
Click an adapter's name, or choose Operational Console from its row menu, to open the console. This is where you diagnose and act on a single connection. It has a Control Panel on the left, a Resource Explorer on the right, a link to this connection's events in the Global Event Log, and a Support Diagnostics download.
Control Panel
The Control Panel shows the live status, the Enabled toggle, the operator actions, and the runtime detail fields. A live indicator next to the title shows when the panel last refreshed; the panel re-polls automatically, and a Refresh Status button forces an update.
The detail fields are the console's runtime facts:
| Field | What it tells you |
|---|---|
| Instance ID | Which adapter-server instance is holding the connection. |
| Version | The adapter version currently running. |
| Mode | Single Writer or Scale Out. |
| Last Heartbeat | When the instance last checked in. A heartbeat that stops advancing is the first signal that the hosting instance has lost contact with the platform. |
| Started | When the current run began. Compare it with Last Heartbeat to judge how long the session has been up and whether it recently restarted. |
The Control Panel also raises a mismatch message when intent and reality disagree: Adapter is disabled but currently running. It should be Shutdown. or Enabled. Waiting for instance to start.... Either message means Admin State and Operational Status do not line up yet.
Taking action from the console
The console exposes an Enabled toggle and three action buttons. They are not interchangeable; match the action to the symptom.
| Action | What it does | Use when |
|---|---|---|
| Enable | Toggle the Enabled switch on. Sets Admin State to enabled; the platform provisions the instance. | The adapter should be running and is not. |
| Disable | Toggle the Enabled switch off. Sets Admin State to disabled. Disabling an instance that is currently running prompts a confirmation, because it stops the instance and drops the live session. | You want the adapter stopped and kept stopped. |
| Resume | Available when the adapter is Paused or Standby. Resumes processing. Applies immediately with no confirmation. | A paused adapter should start processing again. |
| Pause | Available when the adapter is Running. Stops traffic processing while keeping the instance in place. Confirmation required. | You need to hold processing without tearing the instance down. |
| Shutdown | Available unless the adapter is already Inactive. Stops the running instance; it will not come back until re-enabled. Confirmation required. | You need to stop the running instance now. |
| Edit Configuration | Shown when the adapter is disabled. Opens the configuration wizard. | You need to change the connection's configuration. |
Delete is not on the console; it lives on the fleet list, in the row menu and the bulk bar.
Pause and Resume keep the instance in place. Shutdown stops it, and it stays down until you Enable it again. None of these fix a configuration problem: if the adapter is in Error because of its configuration, use Edit Configuration. For a clean stop-and-start of a running instance, use Restart from the fleet list.
Events
The console does not embed the event log. Use View in Global Event Log to open the Global Event Log pre-filtered to this connection. The event log is the single most useful diagnostic: it is the per-connection narrative, newest first.
| Read the event log to | Because |
|---|---|
| Confirm a connection came up after Enable | A logon event proves the session reached the external system. Its absence means the attempt stopped before logon. |
| Understand why messages are not flowing downstream | Entries show when a pipeline stage dropped a message and which filter caused it. |
| Investigate a disconnect | The error or logout entry names the reason the session ended. Entries immediately before it show whether traffic was flowing at the time. |
Filter the log to cut noise during an incident.
| Filter | Use when |
|---|---|
Severity = Error | Something is wrong and you want to skip routine activity. Severities are Info, Warning, and Error. |
Taxonomy = Protocol | A session or message-level problem is suspected. |
Taxonomy = Security or Audit | You are looking at access or operator-command records. |
| Filter by user | An admin-state change (Enable or Disable) did not have the expected effect and you want to see who did it. These are the operator actions recorded with an actor. |
| Start time / End time | The outage started at a known time and you want to read only that window. |
The Global Event Log also lets you copy the current view or export it to CSV for a ticket. For the full list of event types and the mechanical fix for each, see Event Log Reference.
Support Diagnostics
An Experimental one-click download at the bottom of the console packages a support zip ready to attach to a ticket:
| Entry | Contents |
|---|---|
logs.txt | The instance's recent rolling log window, streamed live from the pod. |
config.json | The adapter's resolved configuration. |
perf.json | The last 30 minutes of latency and throughput evidence: pipeline stage decomposition, durability settlement, target dispatch, backpressure, JVM pressure, and network round-trip, each number annotated with exactly what it measures. |
events.json | The adapter's event timeline for the last 7 days (up to 200 entries, newest first), served from the platform's own store rather than the pod. It survives instance failure, so a bundle for a crashed or stopped adapter still records what happened last: disconnects, errors, and test-run completions with their verdicts. |
manifest.json | Provenance: adapter build version, plugin versions the pod actually loaded (live, or as of its last lease claim when the pod is offline), and an explicit list of anything the bundle could not collect. |
The same perf.json payload is available on demand at GET /api/v1/adapters/{adapterId}/perf without downloading a bundle.
See also
- Event Log Reference: how to read and filter the event log, and where event meanings live.
- What Happens to Messages: how the pipeline handles delivery failures and retries.
- Adapters Overview: the meaning of Operational Status and Admin State, and the Single Writer and Scale Out deployment modes.