Running Scribe

Starting a Scribe Server:

scribed [ -p <port number> ] [ -c <configuration file> ]

Stopping a Scribe Server:

scribe_ctrl stop [ <port number> ]

Using the scribe_ctrl script to monitor Scribe:

scribe_ctrl {command} [ <port number> ]

scribe_ctrl commands:
status – Returns ‘ALIVE’ if the server is running and healthy
version – Returns the version of the Scribe Server
alive – Returns the time the server was started (unix time format, in seconds)
stop – Stops the scribe server gracefully
reload – Reloads the config file and reconfigures Scribe
counters – Returns the following statistics (if non-zero):

• received good: Number of messages received since Scribe Server was started
• received bad: Number of invalid messages received
• received_good_bytes: Number of bytes received since Scribe Server was started
• received_bad_bytes: Number of invalid bytes received
• sent: Number of messages sent to another Scribe Server
• denied for queue size: Number of requests denied due to a full message queue
• denied for rate: Number of requests denied due to rate limitting
• retries: Number of times a Buffer Store had to retry logging a batch of messages
• requeue: Number of times Scribe had to retry sending messages to a store (if must_succeed is enabled).
• lost: Number of messages that were not logged. (Recommended configuration: Use BufferStores to avoid lost messages.)
• received blank category: Number of messages received without a message category
• < category >_< fsType >_wrote_num_messages: Number of messages were written to fstype(hdfs/std) since scribe was started. This metric is emitted at thread level also in case of multiple threads
• < category >_< fsType >_wrote_num_bytes: Number of bytes were written to fstype(hdfs/std) since scribe was started. This metric is emitted per thread also in case of multiple writer threads.