VM test: count a game compressed as ogc does, not by every extent #49
Loading…
Add table
Add a link
Reference in a new issue
No description provided.
Delete branch "vm-flake"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
The redownload scenario failed now and then ("game 200 was not compressed" in #47's first run, "game 100" in a local reproduction). The failing file had one plain single-block extent out of 12288. btrfs never compresses a write-back range of one block, and write-back that starts while the fake Steam is still writing onto the compress=zstd mount can leave one behind. ogc counts a file that is at least 98 % compressed as done (
ExtentStats::is_mostly_compressed) and rightly leaves it; the test wanted every extent encoded.compressed()now counts blocks and accepts at most 2 % plain, ogc's own threshold; a file nobody compressed still fails (100 % plain), and the failure message shows the counts. The helpers move totests/vm/common.shsoguest.shstays under 500 lines. Not a product bug: two independent investigations found no case of ogc compressing a game Steam was still writing.Verified: redownload 4× in a row, then redownload, space and convert after the move, all pass.