Issue Summary

When running large or complex migrations using JCMA, one of the most common questions raised during the migration itself or when reviewing a slow or stalled run is: "What is the throughput of this migration?"

Currently, the support logs that JCMA generates contain no summarized throughput data. This means there is no direct way to understand how fast a migration is progressing, how long it is likely to take, or whether performance has degraded — from the logs alone.

This is a recurring gap for any migrator running a large migration where time, data volume, and cutover windows matter.

Steps to Reproduce: 

• Start a JCMA migration
• Check the support.zip - atlassian-jira.log or atlassian-jira-migration.log
• Migration throughput details are not recorded

Expected Results

Structured, summarized throughput entries to the JCMA migration logs so that migrators can understand migration speed and progress directly from the logs — without needing external tooling or support intervention.

What Should Be Logged

1. Per-Phase Throughput Summary — written when each phase completes:

[THROUGHPUT] Phase=EXPORT      Duration=42m   EntitiesProcessed=84,320   Rate=2,008 entities/min   Status=COMPLETED
[THROUGHPUT] Phase=IMPORT      Duration=67m   EntitiesProcessed=84,100   Rate=1,255 entities/min   Status=COMPLETED
[THROUGHPUT] Phase=ATTACHMENTS Duration=180m  DataTransferred=48.2GB     Rate=4.47 MB/s            Status=COMPLETED 

2. Periodic Progress Snapshots — written every ~15 minutes during active migration:

[THROUGHPUT] Snapshot  Phase=ATTACHMENTS  Elapsed=45m  Completed=12,800  Remaining=~36,200  CurrentRate=284 files/min  ETA=~127min 

3. Migration-Level Summary — written on migration completion:
 

[THROUGHPUT] MIGRATION COMPLETE  TotalDuration=289m  TotalEntities=84,320  Attachments=49,000 (48.2GB)
             ExportRate=2,008 ent/min  ImportRate=1,255 ent/min  AttachmentRate=4.47 MB/s 

Actual Results
No throughput summary is written to any log file

Workaround

Currently there is no known workaround for this behavior. A workaround will be added here when available