Tar metadata needs to be generate from Python rather then tar
Registered by
Fausto Marzi
Currently freezer uses tar --incremental feature wrapped called by subprocess.Popen to generate the meta data file to execute incremental backup. Durin some tests, to generate the metadata on ~250k files on a live file system (with lvm snapshot), it took more then 6 hours to generate the meta data. This is a well know issue in tar from 2011. The metadata file needs to be generated from python, possibly using libarchive to generate a gnu/tar compatible meta data file.
Blueprint information
- Status:
- Not started
- Approver:
- None
- Priority:
- High
- Drafter:
- Fausto Marzi
- Direction:
- Needs approval
- Assignee:
- Fausto Marzi
- Definition:
- Discussion
- Series goal:
- None
- Implementation:
- Unknown
- Milestone target:
- None
- Started by
- Completed by
Whiteboard
(?)