Changed drop ordering when using git annex sync --content or the assistant, to drop from remotes first and from the local repo last. This works better with the behavior changes to drop in many cases.

Annex/Drop.hs

@@ -30,7 +30,10 @@ type Reason = String
  - The UUIDs are ones where the content is believed to be present.
  - The Remote list can include other remotes that do not have the content;
  - only ones that match the UUIDs will be dropped from.
- - If allowed to drop fromhere, that drop will be tried first.
+ -
+ - If allowed to drop fromhere, that drop will be done last. This is done
+ - because local drops do not need any LockedCopy evidence, and so dropping
+ - from local last allows the content to be removed from more remotes.
  -
  - A VerifiedCopy can be provided as an optimisation when eg, a key
  - has just been uploaded to a remote.
@@ -52,8 +55,8 @@ handleDropsFrom locs rs reason fromhere key afile preverified runner = do
 		, return $ maybeToList afile
 		)
 	n <- getcopies fs
-	if fromhere && checkcopies n Nothing
-		then go fs rs =<< dropl fs n
+	void $ if fromhere && checkcopies n Nothing
+		then go fs rs n >>= dropl fs
 		else go fs rs n
   where
 	getcopies fs = do
@@ -78,12 +81,12 @@ handleDropsFrom locs rs reason fromhere key afile preverified runner = do
 		| S.member u untrusted = v
 		| otherwise = decrcopies v Nothing
 
-	go _ [] _ = noop
+	go _ [] n = pure n
 	go fs (r:rest) n
 		| uuid r `S.notMember` slocs = go fs rest n
 		| checkcopies n (Just $ Remote.uuid r) =
 			dropr fs r n >>= go fs rest
-		| otherwise = noop
+		| otherwise = pure n
 
 	checkdrop fs n u a
 		| null fs = check $ -- no associated files; unused content

debian/changelog

@@ -14,6 +14,9 @@ git-annex (5.20150931) UNRELEASED; urgency=medium
     - If a git remote has too old a version of git-annex-shell installed,
       git-annex won't trust it to hold onto a copy of a file when dropping
       that file from some other remote.
+  * Changed drop ordering when using git annex sync --content or the
+    assistant, to drop from remotes first and from the local repo last.
+    This works better with the behavior changes to drop in many cases.
   * Do verification of checksums of annex objects downloaded from remotes.
   * When annex objects are received into git repositories from other git
     repos, their checksums are verified then too.
@@ -36,6 +39,7 @@ git-annex (5.20150931) UNRELEASED; urgency=medium
     and stop recommending bittornado | bittorrent.
   * Debian: Remove dependency on transformers library, as it is now
     included in ghc.
+  * 
 
  -- Joey Hess <id@joeyh.name>  Thu, 01 Oct 2015 12:42:56 -0400
 

doc/bugs/concurrent_drop--from_presence_checking_failures.mdwn

@@ -367,3 +367,32 @@ for 1 locked copy when dropping from a remote, and so is sufficent to
 fix the bug.
 
 > done
+
+# drop ordering
+
+Consider the network: `T -- C --- (B1,B2)`
+
+When numcopies is 2, a file could start on T, get copied to C, and then on
+the B1 and B2. At which point, it can be dropped from C and T (if
+unwanted).
+
+Currently, the assistant and sync --content drop from the local repo 1st,
+and then from remotes. Before the changes to fix this bug, that worked;
+the content got removed from T and then from C, using the copies on B1 and
+B2 as evidence. Now though, if B1 and B2 are special remotes, once the copy
+is dropped from C, there is no locked copy available on (B1,B2), so the
+subsequent drop from T fails.
+
+Changing the drop order lets C lock its own copy in order to
+drop from T. then the local drop from C proceeds successfully without locking,
+as local drops don't need locking.
+
+Course, there is a behavior change here.. Before, if B2 didn't exist,
+the content would reach B2 and then be dropped from C, and then with only 2
+copies, it could not also be dropped from T. If the drop order changes, the
+content is instead dropped from T and left on C.
+
+The new behavior seems better when T is a transfer remote, but perhaps not in
+other cases.
+
+> implemented