compact: free space per pack with --threshold#9801
Draft
mr-raj12 wants to merge 1 commit into
Draft
Conversation
Group objects by their pack and act per pack: drop fully-unused packs, and rewrite mixed packs whose unused bytes reach --threshold (default 40%) by copying the survivors forward with compact_pack. Two index scans keep the memory use bounded. refs borgbackup#8572 borgbackup#8514
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## master #9801 +/- ##
==========================================
- Coverage 84.77% 81.75% -3.02%
==========================================
Files 92 92
Lines 15251 15284 +33
Branches 2286 2295 +9
==========================================
- Hits 12929 12496 -433
- Misses 1622 2091 +469
+ Partials 700 697 -3 ☔ View full report in Codecov by Harness. |
ThomasWaldmann
requested changes
Jun 23, 2026
| pack_total, pack_unused = {}, {} | ||
| for id, entry in self.chunks.iteritems(): | ||
| pid = entry.pack_id | ||
| pack_total[pid] = pack_total.get(pid, 0) + entry.obj_size |
Member
There was a problem hiding this comment.
this is what defaultdict type was made for.
| continue # all used -> leave alone | ||
| if unused == total: | ||
| drop_packs.add(pid) # all unused -> drop the whole file | ||
| elif unused / total * 100 >= self.threshold: |
Member
There was a problem hiding this comment.
100 * unused / total is simpler to read.
Comment on lines
+329
to
+336
| subparser.add_argument( | ||
| "--threshold", | ||
| metavar="PERCENT", | ||
| dest="threshold", | ||
| type=float, | ||
| default=40.0, | ||
| help="rewrite a pack when at least PERCENT of its bytes are unused (default: 40)", | ||
| ) |
Member
There was a problem hiding this comment.
compare cli syntax and default value to what we had in borg 1.4-maint.
Member
There was a problem hiding this comment.
1.4 used an int cli option and defaulted to 10%.
|
|
||
| repository = Repository(location, exclusive=True, create=True) | ||
| build_one_pack(repository, wasteful) | ||
| repository = reopen(repository) |
Member
There was a problem hiding this comment.
didn't we find already that we don't really need reopen?
Member
|
Also please rebase on current master (I have merged #9800 now). |
Member
|
Have a look at that, maybe you can implement |
Member
|
I did a test run with slightly modified code (N=2) and got this. So guess the 100% display issue is not fixed yet? Also: even without -v, it is very verbose: I suggest you try this manually to get a feel of it and find such issues yourself. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Description
Moves
borg compactfrom deleting single objects to compacting whole packs, so it keeps working once a pack holds more than one object (N>1).For each pack:
--thresholdpercent (default 40), copying the survivors into a new pack viaRepository.compact_packand dropping the old one. Below the threshold the pack is left alone, so a large pack is not rewritten just to reclaim a few bytes.The chunk index is scanned twice to keep memory bounded: first only per-pack byte counts to decide each pack's fate, then the object ids of just the packs that change. The #9748 crash-safety order is preserved: cached chunk indexes are invalidated before the first store change.
At N=1 every pack holds one object, so mixed packs never occur and the behavior matches before. The rewrite path is covered by a test that forces max_count > 1.
This recycles the approach from #9777, which can be closed.
refs #8572 #8514
Checklist
master