SERVER-44017 hang_analyzer.py inflates debug-symbols

2 files changed, 61 insertions, 26 deletions

diff --git a/buildscripts/hang_analyzer.py b/buildscripts/hang_analyzer.py
index 67a4ef97e1c..ce92c7b1be5 100755
--- a/buildscripts/hang_analyzer.py
+++ b/buildscripts/hang_analyzer.py
@@ -487,6 +487,61 @@ class JstackWindowsDumper(object):
         root_logger.warning("Debugger jstack not supported, skipping dumping of %d", pid)
 
 
+class DebugExtractor(object):
+    """Extracts `mongo-debugsymbols.tgz`."""
+
+    @staticmethod
+    def extract_debug_symbols(root_logger):
+        """
+        Extract debug symbols. Idempotent.
+
+        :param root_logger: logger to use
+        :return: None
+        """
+        path = os.path.join(os.getcwd(), 'mongo-debugsymbols.tgz')
+        root_logger.debug('Starting: Extract debug-symbols from %s.', path)
+        if not os.path.exists(path):
+            root_logger.info('Debug-symbols archive-file does not exist. '
+                             'Hang-Analyzer may not complete successfully, '
+                             'or debug-symbols may already be extracted.')
+            return
+        try:
+            DebugExtractor._exxtract_tar(path, root_logger)
+            root_logger.debug('Finished: Extract debug-symbols from %s.', path)
+        # We never want this to cause the whole task to fail.
+        # The rest of hang_analyzer.py will continue to work without the
+        # symbols it just won't be quite as helpful.
+        # pylint: disable=broad-except
+        except Exception as exception:
+            root_logger.warning('Error when extracting %s: %s', path, exception)
+
+    @staticmethod
+    def _exxtract_tar(path, root_logger):
+        import shutil
+        # The file name is always .tgz but it's "secretly" a zip file on Windows :(
+        compressed_format = 'zip' if _IS_WINDOWS else 'gztar'
+        shutil.unpack_archive(path, format=compressed_format)
+        for (src, dest) in DebugExtractor._extracted_files_to_copy():
+            if os.path.exists(dest):
+                root_logger.debug('Debug symbol %s already exists, not copying from %s.', dest, src)
+                continue
+            shutil.copy(src, dest)
+            root_logger.debug('Copied debug symbol %s.', dest)
+
+    @staticmethod
+    def _extracted_files_to_copy():
+        out = []
+        for ext in ['debug', 'dSYM', 'pdb']:
+            for file in ['mongo', 'mongod', 'mongos']:
+                # need to glob because it untar's to a directory that looks like
+                # mongodb-linux-x86_64-enterprise-rhel62-4.3.0-1823-gb9c13fa-patch-5daa05630ae60652f0890f76
+                haystack = os.path.join('mongodb*', 'bin', '{file}.{ext}'.format(
+                    file=file, ext=ext))
+                for needle in glob.glob(haystack):
+                    out.append((needle, os.path.join(os.getcwd(), os.path.basename(needle))))
+        return out
+
+
 def get_hang_analyzers():
     """Return hang analyzers."""
 
@@ -609,6 +664,8 @@ def main():  # pylint: disable=too-many-branches,too-many-locals,too-many-statem
     except AttributeError:
         root_logger.warning("Cannot determine Unix Current Login, not supported on Windows")
 
+    DebugExtractor.extract_debug_symbols(root_logger)
+
     interesting_processes = ["mongo", "mongod", "mongos", "_test", "dbtest", "python", "java"]
     go_processes = []
     process_ids = []
diff --git a/etc/evergreen.yml b/etc/evergreen.yml
index 2f2c96dca1d..9010fbb2a60 100644
--- a/etc/evergreen.yml
+++ b/etc/evergreen.yml
@@ -48,6 +48,8 @@ variables:
   - name: compile
   commands:
   - func: "do setup"
+  - func: "set task expansion macros"
+  - func: "fetch debugsymbols archive"
   - func: "run tests"
     vars:
       resmoke_args: --help
@@ -815,30 +817,6 @@ functions:
       bucket: mciuploads
       local_file: src/mongo-debugsymbols.tgz
 
-  "extract debugsymbols": &extract_debugsymbols
-    command: shell.exec
-    params:
-      working_dir: src
-      script: |
-        set -o errexit
-        set -o verbose
-        # Debug symbols are not created for every variant
-        if [ ! -f mongo-debugsymbols.tgz ]; then
-          exit
-        fi
-        ${decompress} mongo-debugsymbols.tgz
-        files="mongo mongod mongos"
-        file_exts="debug dSYM pdb"
-        for file_ext in $file_exts
-        do
-            for file in $files
-            do
-                mv mongodb*/bin/$file.$file_ext . 2>/dev/null || true
-            done
-            rm -r mongodb*/bin/*.$file_ext 2>/dev/null || true
-        done
-        rm mongo-debugsymbols.tgz 2>/dev/null || true
-
   "fetch and build OpenSSL":
     command: shell.exec
     params:
@@ -1488,7 +1466,6 @@ functions:
           path_value="$path_value:${task_path_suffix}"
         fi
 
-
         # The "resmoke_wrapper" expansion is used by the 'burn_in_tests' task to wrap the resmoke.py
         # invocation. It doesn't set any environment variables and should therefore come last in
         # this list of expansions.
@@ -1657,6 +1634,8 @@ functions:
         value: mongodata_aws
       - key: aws_secret_remote
         value: ${mongodatafiles_aws_secret}
+  - *set_task_expansion_macros
+  - *fetch_debugsymbols_archive
   - *set_up_remote_credentials
   - *determine_resmoke_jobs
   - *update_resmoke_jobs_expansions
@@ -3647,7 +3626,6 @@ post:
 # Timeout steps
 timeout:
   - func: "fetch debugsymbols archive"
-  - func: "extract debugsymbols"
   - func: "get EC2 address"
   - func: "update EC2 address"
   - func: "run hang analyzer"