capture error for mapping (#5325)

fix b/524480139

This PR update the batch service to handle when the config mapping
doesn't match. The problem is that this raises a exception which avoids
to ack the message from pubsub, keeping the problematic task in the
queue forever. For this case of mismatching, we should log and remove
the task from the queue to avoid no scheduling legit tasks.

Signed-off-by: Javan Lacerda <javanlacerda@google.com>

Author: javanlacerda

src/clusterfuzz/_internal/batch/service.py

@@ -356,8 +356,15 @@ def _get_specs_from_config(
     if (task.command, task.job_type) in specs:
       # Don't repeat work for no reason.
       continue
-    config_name, disk_size_gb, base_os_version = config_map[(task.command,
-                                                             task.job_type)]
+    try:
+      config_name, disk_size_gb, base_os_version = config_map[(task.command,
+                                                               task.job_type)]
+    except Exception:
+      logs.error(
+          'Error on mapping the job',
+          command=task.command,
+          job_type=task.job_type)
+      continue
 
     instance_spec = batch_config.get('mapping').get(config_name)
     if instance_spec is None: