Description
We've experienced at least a few fatal errors in ChangelLogEntryTemp processing, and while the Daemon-log messages are helpful, each one becomes an archeological dig because the problematic entry can be anywhere in the first commit-batch of entries, and the available information (group id or whatever) matches different CLET columns depending on the event type. For instance, here is what is being displayed on a recent failure:
java.lang.RuntimeException: Active PITGroup with sourceId=4b3b35de1931406c83a644539b985428 not found,
The fact that this problem happens to be in the string02 column is far from clear to us.
This Jira is suggesting two improvements:
- Log some unique CLET information that leads to the specific problem. From the table's indexes, it looks like this is probably the 'id' column
- When an Exception is aborting CLET processing, repeat the processing with a commit-batch size of 1 so the oldest entry that is left is the one causing the problems.