Define a max txlog size for which no more txlog merges are attempted.
This would prevent consequential errors when txlogs get too large and make the backup container more stable in general.
Observed consequential error:
- Failing txlog merge is retried over and over again, producing many useless parts on s3.
Increases also stability because:
- During the txlog merge, the execution of other tasks (such as uploading WAL segments or creating new base backups) may be blocked.
Make the threshold txlog size configurable!
Edited by Patrik Dürrenberger