Details
-
Improvement
-
Resolution: Fixed
-
Major
-
6.6.1
-
1
Description
Today, when cbcollect_info runs, it first generates all of the files that it wishes to store in the cbcollect_info and puts these in the temporary location.
After that, if redaction is enabled, it will make a redacted copy of every single file and store this in a /redacted sub-directory.
At this stage, we're using twice as much disk space as the total uncompressed files within the cbcollect!
Only after this completes do we then move these into zip files.
Instead to keep the disk footprint for cbcollect_info as lean as possible, we should do the following:
- Move each individual file into the output zip as the generation finishes
- Redact each file (if applicable) just before moving into the output zip
This would ensure that we're not wasting potentially huge amounts (over 30GB has been seen in the field) when collecting logs.