Closed rzvoncek closed 8 months ago
Attention: Patch coverage is 97.34513%
with 3 lines
in your changes are missing coverage. Please review.
Project coverage is 80.02%. Comparing base (
49f88ea
) to head (64f9eb6
).
Finalised the reimplementation to not even look at the manifests.
DId a basic manual test with corrupting a file (rm
then touch
it).
Pushed a commit with the fixes. Also rebased on master to fix merge conflicts.
Issues
4 New issues
0 Accepted issues
Measures
0 Security Hotspots
No data about Coverage
0.0% Duplication on New Code
Fixes #709 Fixes #368
In short, the way we do this is to list all files in storage, in the data folder for the given prefix and node, after doing a snapshot but before we start backing up individual tables. We we then group the listed files into a dict->dict->set keyed by keyspace->table, so that we can look up any given file efficiently.
Tested manually on a node with ~11k LCS SSTables. Did not observe a particular increase in (differential) backup duration. There probably is an increased memory consumption, but I'm struggling a bit to quantify how much.