From bf2b9c5e9ad5568321fd9652ef59b85951bca963 Mon Sep 17 00:00:00 2001
From: Corentin Hembise <corentin@lalilo.com>
Date: Sat, 5 Dec 2020 19:31:07 +0100
Subject: [PATCH 1/2] Add context about failing branches during error in
 used_cache()

---
 dvc/repo/__init__.py | 34 +++++++++++++++++++---------------
 1 file changed, 19 insertions(+), 15 deletions(-)

diff --git a/dvc/repo/__init__.py b/dvc/repo/__init__.py
index 19b1e4f505..2ab7c3fefe 100644
--- a/dvc/repo/__init__.py
+++ b/dvc/repo/__init__.py
@@ -8,7 +8,7 @@
 from git import InvalidGitRepositoryError
 
 from dvc.config import Config
-from dvc.exceptions import FileMissingError
+from dvc.exceptions import DvcException, FileMissingError
 from dvc.exceptions import IsADirectoryError as DvcIsADirectoryError
 from dvc.exceptions import NotDvcRepoError, OutputNotFoundError
 from dvc.path_info import PathInfo
@@ -319,22 +319,26 @@ def used_cache(
         ):
             targets = targets or [None]
 
-            pairs = cat(
-                self.stage.collect_granular(
-                    target, recursive=recursive, with_deps=with_deps
+            try:
+                pairs = cat(
+                    self.stage.collect_granular(
+                        target, recursive=recursive, with_deps=with_deps
+                    )
+                    for target in targets
                 )
-                for target in targets
-            )
 
-            suffix = f"({branch})" if branch else ""
-            for stage, filter_info in pairs:
-                used_cache = stage.get_used_cache(
-                    remote=remote,
-                    force=force,
-                    jobs=jobs,
-                    filter_info=filter_info,
-                )
-                cache.update(used_cache, suffix=suffix)
+                suffix = f"({branch})" if branch else ""
+                for stage, filter_info in pairs:
+                    used_cache = stage.get_used_cache(
+                        remote=remote,
+                        force=force,
+                        jobs=jobs,
+                        filter_info=filter_info,
+                    )
+                    cache.update(used_cache, suffix=suffix)
+            except DvcException as e:
+                logger.error(f"Cache for '{branch}' could not be collected")
+                raise e
 
         if used_run_cache:
             used_cache = self.stage_cache.get_used_cache(

From eb532efd99085ff212aa0e18902d2fd28a1f1ecf Mon Sep 17 00:00:00 2001
From: Corentin Hembise <corentin@lalilo.com>
Date: Sat, 5 Dec 2020 19:32:03 +0100
Subject: [PATCH 2/2] Add ability to ignore dvc exceptions during used_cache()

---
 dvc/command/gc.py    | 11 +++++++++++
 dvc/repo/__init__.py |  5 +++++
 dvc/repo/gc.py       |  2 ++
 3 files changed, 18 insertions(+)

diff --git a/dvc/command/gc.py b/dvc/command/gc.py
index 2948da26b7..7c2bf8c1e0 100644
--- a/dvc/command/gc.py
+++ b/dvc/command/gc.py
@@ -55,6 +55,7 @@ def run(self):
             jobs=self.args.jobs,
             repos=self.args.repos,
             workspace=self.args.workspace,
+            skip_failing_collect=self.args.skip_failing_collect,
         )
         return 0
 
@@ -141,4 +142,14 @@ def add_parser(subparsers, parent_parser):
         "Useful if you share a single cache across repos.",
         metavar="<paths>",
     )
+    gc_parser.add_argument(
+        "--skip-failing-collect",
+        action="store_true",
+        default=False,
+        help=(
+            "Skip commits were the cache cannot be "
+            "collected due to user errors in the dvc "
+            "state (eg. badly formatted dvc.yaml)."
+        ),
+    )
     gc_parser.set_defaults(func=CmdGC)
diff --git a/dvc/repo/__init__.py b/dvc/repo/__init__.py
index 2ab7c3fefe..d82d62cac4 100644
--- a/dvc/repo/__init__.py
+++ b/dvc/repo/__init__.py
@@ -293,6 +293,7 @@ def used_cache(
         jobs=None,
         recursive=False,
         used_run_cache=None,
+        ignore_dvc_exceptions=False,
     ):
         """Get the stages related to the given target and collect
         the `info` of its outputs.
@@ -338,6 +339,10 @@ def used_cache(
                     cache.update(used_cache, suffix=suffix)
             except DvcException as e:
                 logger.error(f"Cache for '{branch}' could not be collected")
+                if ignore_dvc_exceptions:
+                    logger.debug(e)
+                    continue
+
                 raise e
 
         if used_run_cache:
diff --git a/dvc/repo/gc.py b/dvc/repo/gc.py
index 9b415b7a87..cf7c285222 100644
--- a/dvc/repo/gc.py
+++ b/dvc/repo/gc.py
@@ -30,6 +30,7 @@ def gc(
     jobs=None,
     repos=None,
     workspace=False,
+    skip_failing_collect=False,
 ):
 
     # require `workspace` to be true to come into effect.
@@ -66,6 +67,7 @@ def gc(
                     remote=remote,
                     force=force,
                     jobs=jobs,
+                    ignore_dvc_exceptions=skip_failing_collect,
                 )
             )