PR #1824: clean up images that erroneously think they’re in git

Author: reidpr

lib/build_cache.py

@@ -1324,14 +1324,28 @@ def worktrees_fix(self):
  // "*" // im.GIT_DIR)) }
  wt_gits = { fs.Path(i).name
  for i in glob.iglob("%s/worktrees/*" % self.root) }
+ # Unlink images that think they are in Git but are not. This should not
+ # happen, but it does, and I wasn’t able to figure out how it happened.
+ wt_gits_orphaned = wt_actuals - wt_gits
+ for img_dir in wt_gits_orphaned:
+ link = ch.storage.unpack_base // img_dir // im.GIT_DIR
+ ch.WARNING("image erroneously marked cached, fixing: %s" % link,
+ ch.BUG_REPORT_PLZ)
+ link.unlink()
+ wt_actuals -= wt_gits_orphaned
  # Delete worktree data for images that no longer exist or aren’t
  # Git-enabled any more.
  wt_gits_deleted = wt_gits - wt_actuals
  for wt in wt_gits_deleted:
  (ch.storage.build_cache // "worktrees" // wt).rmtree()
  ch.VERBOSE("deleted %d stale worktree metadatas" % len(wt_gits_deleted))
  wt_gits -= wt_gits_deleted
- assert (wt_gits == wt_actuals)
+ # Validate that the pointers are in sync now.
+ if (wt_gits != wt_actuals):
+ ch.ERROR("found images -> cache links: %s" % " ".join(wt_actuals))
+ ch.ERROR("found cache -> images links: %s" % " ".join(wt_gits))
+ ch.FATAL("build cache is desynchronized, cannot proceed",
+ ch.BUG_REPORT_PLZ)
  # If storage directory moved, repair all the paths.
  if (len(wt_gits) > 0):
  wt_dir_stored = fs.Path(( ch.storage.build_cache

test/build/55_cache.bats

@@ -1100,6 +1100,7 @@ EOF
  diff -u <(echo "$blessed_out") <(echo "$output" | treeonly)
 }
 
+
 @test "${tag}: multistage COPY" {
  # Multi-stage build with no instructions in the first stage.
  df_no=$(cat <<'EOF'
@@ -1502,6 +1503,7 @@ EOF
  [[ ! -e $CH_IMAGE_STORAGE/img/tmpimg/var/lib/rpm/__db.001 ]]
 }
 
+
 @test "${tag}: restore ACLs, xattrs" { # issue #1287
  # Check if test needs to be skipped
  touch "$BATS_TMPDIR/tmpfs_test"
@@ -1538,3 +1540,39 @@ EOF
  [[ $status -eq 0 ]]
  [[ $output = *"user:$USER:r--"* ]]
 }
+
+
+@test "${tag}: orphaned worktrees" { # PR #1824
+ img_metadata=$CH_IMAGE_STORAGE/img/tmpimg/ch
+ img_to_git=$img_metadata/git
+ git_worktrees=$CH_IMAGE_STORAGE/bucache/worktrees
+ git_to_img=$git_worktrees/tmpimg
+
+ # pull image, should be unlinked
+ ch-image pull --no-cache scratch tmpimg
+ ch-image build-cache # rm leftover $git_to_img if it exists
+ ls -lh "$img_metadata" "$git_worktrees"
+ [[ ! -e "$img_to_git" ]]
+ [[ ! -e "$git_to_img" ]]
+
+ # add fake link
+ touch "$img_to_git"
+ ls -lh "$img_metadata" "$git_worktrees"
+ [[ -e "$img_to_git" ]]
+ [[ ! -e "$git_to_img" ]]
+
+ # ch-image should warn and fix instead of crashing
+ run ch-image list
+ echo "$output"
+ [[ $status -eq 0 ]]
+ [[ $output = *'image erroneously marked cached, fixing'* ]]
+
+ # warning should now be gone and the state be good
+ ls -lh "$img_metadata" "$git_worktrees"
+ [[ ! -e "$img_to_git" ]]
+ [[ ! -e "$git_to_img" ]]
+ run ch-image list
+ echo "$output"
+ [[ $status -eq 0 ]]
+ [[ $output != *'image erroneously marked cached, fixing'* ]]
+}