Analyze Transaction Replay Performance

Use the analyze_replay_performance.py script after a database initialization completes to analyze transaction replay throughput statistics, including the replay duration, any throttling, the number of transactions replayed, and the number of transactions replayed per second.

About this task

The analyze_replay_performance.py script analyzes the checkpoint or checkpoints in the mtx_debug.log file on the active publishing server to analyze the replay performance. This server drives the replay locally and remotely. If you run the script and an error occurs stating a replay item could not be found, the mtx_debug.log file was closed before the entire checkpoint was written. In such cases, you must specify the first mtx_debug.log file and the next sequential one.

Perform this task on the processing server of the active engine that serviced the replay request. Look for database initialization requester or start sending in-memory database replay to peer cluster log messages to find the correct processing server.

Procedure

Run the analyze_replay_performance.py script to analyze the transaction replay statistics written to the mtx_debug.log file. To print more information about the replay performance, you can specify the -v (verbose) option.

If the mtx_debug.log file has more than one checkpoint and you only want to analyze one, run the analyze_replay_performance.py script with the -c option and specify the checkpoint name.

Note: If a transaction record from a transaction log file in a checkpoint operation or a real-time replay operation is retried for more than a maximum number of times and fails, the Transaction Server logs a critical error and a Creating a possibly in-consistent checkpoint warning and continues processing.