Merge pull request #33 from bento-platform/stream-cache

Add cache control headers to run streams

bento_wes/app.py

@@ -68,7 +68,7 @@ def service_info():
         "description": "Workflow execution service for a CHORD application.",
         "organization": {
             "name": "C3G",
-            "url": "http://www.computationalgenomics.ca"
+            "url": "https://www.computationalgenomics.ca"
         },
         "contactUrl": "mailto:[email protected]",
         "version": bento_wes.__version__,
@@ -87,16 +87,19 @@ def service_info():
         if res_tag := subprocess.check_output(["git", "describe", "--tags", "--abbrev=0"]):
             res_tag_str = res_tag.decode().rstrip()
             info["git_tag"] = res_tag_str
+            # noinspection PyTypeChecker
             info["bento"]["gitTag"] = res_tag_str
         if res_branch := subprocess.check_output(["git", "branch", "--show-current"]):
             res_branch_str = res_branch.decode().rstrip()
             info["git_branch"] = res_branch_str
+            # noinspection PyTypeChecker
             info["bento"]["gitBranch"] = res_branch_str
         if res_commit := subprocess.check_output(["git", "rev-parse", "HEAD"]):
             res_commit_str = res_commit.decode().rstrip()
+            # noinspection PyTypeChecker
             info["bento"]["gitCommit"] = res_commit_str
     except Exception as e:
         except_name = type(e).__name__
-        current_app.logger.error(f"Error retrieving git information: {str(except_name)}: {e}")
+        current_app.logger.info(f"Could not retrieve git information: {str(except_name)}: {e}")
 
     return jsonify(info)

bento_wes/runs.py

@@ -336,8 +336,19 @@ def run_detail(run_id):
 def get_stream(c, stream, run_id):
     c.execute("SELECT * FROM runs AS r, run_logs AS rl WHERE r.id = ? AND r.run_log = rl.id", (str(run_id),))
     run = c.fetchone()
-    return (current_app.response_class(response=run[stream], mimetype="text/plain", status=200) if run is not None
-            else flask_not_found_error(f"Stream {stream} not found for run {run_id}"))
+    return (current_app.response_class(
+        headers={
+            # If we've finished, we allow long-term (24h) caching of the stdout/stderr responses.
+            # Otherwise, no caching allowed!
+            "Cache-Control": (
+                "private, max-age=86400" if run["state"] in states.TERMINATED_STATES
+                else "no-cache, no-store, must-revalidate, max-age=0"
+            ),
+        },
+        response=run[stream],
+        mimetype="text/plain",
+        status=200,
+    ) if run is not None else flask_not_found_error(f"Stream {stream} not found for run {run_id}"))
 
 
 @bp_runs.route("/runs/<uuid:run_id>/stdout", methods=["GET"])

bento_wes/states.py

@@ -13,3 +13,4 @@
 
 FAILURE_STATES = frozenset({STATE_EXECUTOR_ERROR, STATE_SYSTEM_ERROR})
 SUCCESS_STATES = frozenset({STATE_COMPLETE})
+TERMINATED_STATES = FAILURE_STATES.union(SUCCESS_STATES)