Periodic Logging of Server Status

Edit online

The Universal Messaging server writes status information to the log file at regular intervals. The default interval can be configured using the StatusBroadcast realm configuration property, and the default value is 5 seconds).

For information on realm configuration properties, see the section Realm Configuration in the Enterprise Manager part of the Administration Guide.

Here is a sample status log message:


ServerStatusLog> Memory=481, Direct=355, Mapped=0, OpenFileDesc=81, 
EventMemory=0, Disk=81913, CPU=1.5, Scheduled=51, Queued=0, 
Connections=2, BytesIn=3898, BytesOut=194852, Published=0, Consumed=0, 
QueueSize=0, ClientsSize=0, CommQueueSize=0

The following table describes the status log metrics:

Metric	Description	Unit
Memory	Free JVM memory	Megabyte
Direct	Free direct memory	Megabyte
Mapped	Used mapped memory	Megabyte
OpenFileDesc	On Unix only. The number of file descriptors opened by the Universal Messaging server process
EventMemory	JVM memory consumed by events	Megabyte
Disk	Free disk space in UM data directory	Megabyte
CPU	Recent CPU usage for the JVM process in percent. Valid values are from `0` to `100` percent.	%
Scheduled	Number of tasks in the internal scheduler
Queued	Number of total tasks currently queued in all thread pools. A value of `0` indicates that at this moment in time there were no tasks waiting for a thread.
Connections	Active Universal Messaging connections
BytesIn	Total bytes received for currently active Universal Messaging connections	byte
BytesOut	Total bytes sent for currently active Universal Messaging connections	byte
Published	Total number of published events
Consumed	Total number of consumed events
QueueSize	The count of requests/responses awaiting processing that have arrived from other nodes in the cluster (inbound from the cluster). On the master node: the requests are from slave nodes, which forward requests from their locally attached clients to the master node. the responses are from slave nodes, to confirm that the slave nodes have processed client requests previously propagated by the master node. On slave nodes: the requests are from the master node, which propagates (a) requests from clients attached locally to the master, and (b) requests received by the master from clients attached to slave nodes. The propagation goes to all slave nodes in the cluster. the responses are from the master node, to confirm that the master node has processed client requests that were originally sent by the current slave node. A continuously growing size of this queue indicates that the server is not able to process the cluster messages at the rate they are arriving. This value is only present if the node is part of a cluster.
ClientsSize	The count of local client requests awaiting processing by the current node. On the master node, requests from locally attached clients will be processed on the master node, then propagated to the slave nodes in the cluster. On slave nodes, requests from locally attached clients will NOT be processed locally but instead will be forwarded to the master for processing. A continuously growing value of this metric indicates that the server is not able to process and send the client requests to other cluster nodes at the rate these requests are received. This value is only present if the node is part of a cluster.
CommQueueSize	The count of all requests/responses waiting to be sent to the other nodes of the cluster. A continuously growing value of this metric indicates that writing data to the other cluster nodes over the network is slower than the rate at which messages are processed and queued. This indicates a network problem at the data layer level rather than a processing problem on the node. This value is only present if the node is part of a cluster.

For a diagram showing how ClientsSize, QueueSize and CommQueueSize interrelate, see the section Message Passing.

During server startup and each time the log file is rolled, a status log legend is printed in the log file:

ServerStatusLog> Activating Realm status log with format:
Memory - free memory (MB)
Direct - free direct memory (MB)
Mapped - Used mapped memory (MB)
OpenFileDesc - File descriptors opened by process (Unix only)
EventMemory - used event memory (MB)
Disk - free disk space (MB) in server data directory
CPU - cpu load average
Scheduled - scheduled tasks
Queued - total queued tasks in all thread pools
Connections - active connections count
BytesIn - total bytes received (for all active connections)
BytesOut - total bytes send (for all active connections)
Published - total count of published events
Consumed - total count of consumed events
QueueSize - cluster queue size
ClientsSize - cluster client request queue size
CommQueueSize - cluster communication queue size

Periodic logging of the server status can be disabled using the EnableStatusLog realm configuration property.

For information on realm configuration properties, see the section Realm Configuration in the Enterprise Manager part of the Administration Guide.

For information on log rolling, see the section The Enterprise Manager Logs Panel in the Enterprise Manager part of the Administration Guide.