diff --git a/app/main/views/send.py b/app/main/views/send.py
index 8bb6ce24c..bb63f10a9 100644
--- a/app/main/views/send.py
+++ b/app/main/views/send.py
@@ -51,7 +51,6 @@ from app.utils.templates import get_template
 from app.utils.user import user_has_permissions
 from notifications_utils import SMS_CHAR_COUNT_LIMIT
 from notifications_utils.insensitive_dict import InsensitiveDict
-from notifications_utils.logging import scrub
 from notifications_utils.recipients import RecipientCSV, first_column_headings
 from notifications_utils.sanitise_text import SanitiseASCII
 
@@ -953,9 +952,6 @@ def send_notification(service_id, template_id):
             )
         )
 
-    current_app.logger.info(
-        hilite(scrub(f"Recipient for the one-off will be {recipient}"))
-    )
     keys = []
     values = []
     for k, v in session["placeholders"].items():
@@ -971,6 +967,12 @@ def send_notification(service_id, template_id):
     )
     my_data = {"filename": filename, "template_id": template_id, "data": data}
     upload_id = s3upload(service_id, my_data)
+
+    # To debug messages that the user reports have not been sent, we log
+    # the csv filename and the job id.  The user will give us the file name,
+    # so we can search on that to obtain the job id, which we can use elsewhere
+    # on the API side to find out what happens to the message.
+    current_app.logger.info(hilite(f"One-off file: {filename} job_id: {upload_id}"))
     form = CsvUploadForm()
     form.file.data = my_data
     form.file.name = filename
@@ -989,19 +991,6 @@ def send_notification(service_id, template_id):
         valid="True",
     )
 
-    # Here we are attempting to cleverly link the job id to the one-off recipient
-    # If we know the partial phone number of the recipient, we can search
-    # on that initially and find this, which will give us the job_id
-    # And once we know the job_id, we can search on that and it might tell us something
-    # about report generation.
-    current_app.logger.info(
-        hilite(
-            scrub(
-                f"Created job to send one-off, recipient is {recipient}, job_id is {upload_id}"
-            )
-        )
-    )
-
     session.pop("recipient")
     session.pop("placeholders")
 
@@ -1033,7 +1022,12 @@ def send_notification(service_id, template_id):
                 job_id=upload_id,
             )
         )
-
+    total = notifications["total"]
+    current_app.logger.info(
+        hilite(
+            f"job_id: {upload_id} has notifications: {total} and attempts: {attempts}"
+        )
+    )
     return redirect(
         url_for(
             ".view_job",
diff --git a/app/utils/csv.py b/app/utils/csv.py
index 5c5b794de..4ed6d16b5 100644
--- a/app/utils/csv.py
+++ b/app/utils/csv.py
@@ -7,7 +7,6 @@ from flask_login import current_user
 from app.models.spreadsheet import Spreadsheet
 from app.utils import hilite
 from app.utils.templates import get_sample_template
-from notifications_utils.logging import scrub
 from notifications_utils.recipients import RecipientCSV
 
 
@@ -74,12 +73,11 @@ def generate_notifications_csv(**kwargs):
 
     # This generates the "batch" csv report
     if kwargs.get("job_id"):
-        # The kwargs contain the job id, which is linked to the recipient's partial phone number in other debug
         # Some unit tests are mocking the kwargs and turning them into a function instead of dict,
         # hence the try/except.
         try:
             current_app.logger.info(
-                hilite(f"Setting up report with kwargs {scrub(json.dumps(kwargs))}")
+                hilite(f"Setting up report with kwargs {json.dumps(kwargs)}")
             )
         except TypeError:
             pass
@@ -89,7 +87,7 @@ def generate_notifications_csv(**kwargs):
         # we display to 999 characters, because we don't want to show the contents for reports with thousands of rows.
         current_app.logger.info(
             hilite(
-                f"Original csv for job_id {kwargs['job_id']}: {scrub(original_file_contents[0:999])}"
+                f"Original csv for job_id {kwargs['job_id']}: {original_file_contents[0:999]}"
             )
         )
         original_upload = RecipientCSV(
diff --git a/deploy-config/production.yml b/deploy-config/production.yml
index 18b5cfdcc..b38094f28 100644
--- a/deploy-config/production.yml
+++ b/deploy-config/production.yml
@@ -1,6 +1,6 @@
 env: production
 instances: 2
-memory: 1G
+memory: 2G
 public_admin_route: beta.notify.gov
 cloud_dot_gov_route: notify.app.cloud.gov
 redis_enabled: 1
diff --git a/notifications_utils/logging.py b/notifications_utils/logging.py
index 7c56a00ad..4abe5bcf7 100644
--- a/notifications_utils/logging.py
+++ b/notifications_utils/logging.py
@@ -70,6 +70,7 @@ def configure_handler(handler, app, formatter):
     handler.addFilter(AppNameFilter(app.config["NOTIFY_APP_NAME"]))
     handler.addFilter(RequestIdFilter())
     handler.addFilter(ServiceIdFilter())
+    handler.addFilter(PIIFilter())
 
     return handler
 
@@ -134,13 +135,30 @@ class JSONFormatter(BaseJSONFormatter):
         return log_record
 
 
-def scrub(msg):
-    # Eventually we want to scrub all messages in all logs for phone numbers
-    # and email addresses, masking them.  Ultimately this will probably get
-    # refactored into a 'SafeLogger' subclass or something, but let's start here
-    # with phones.
-    phones = re.findall("(?:\\+ *)?\\d[\\d\\- ]{7,}\\d", msg)
-    phones = [phone.replace("-", "").replace(" ", "") for phone in phones]
-    for phone in phones:
-        msg = msg.replace(phone, f"1XXXXX{phone[-5:]}")
-    return msg
+class PIIFilter(logging.Filter):
+    def scrub(self, msg):
+        # Eventually we want to scrub all messages in all logs for phone numbers
+        # and email addresses, masking them.  Ultimately this will probably get
+        # refactored into a 'SafeLogger' subclass or something, but let's start here
+        # with phones.
+
+        # Sometimes just an exception object is passed in for the message, skip those.
+        if not isinstance(msg, str):
+            return msg
+        phones = re.findall("(?:\\+ *)?\\d[\\d\\- ]{7,}\\d", msg)
+        phones = [phone.replace("-", "").replace(" ", "") for phone in phones]
+        for phone in phones:
+            msg = msg.replace(phone, "1XXXXXXXXXX")
+
+        emails = re.findall(
+            r"[\w\.-]+@[\w\.-]+", msg
+        )  # ['alice@google.com', 'bob@abc.com']
+        for email in emails:
+            # do something with each found email string
+            masked_email = "XXXXX@XXXXXXX"
+            msg = msg.replace(email, masked_email)
+        return msg
+
+    def filter(self, record):
+        record.msg = self.scrub(record.msg)
+        return record
diff --git a/tests/notifications_utils/test_logging.py b/tests/notifications_utils/test_logging.py
index 858b9352b..767c1f6f8 100644
--- a/tests/notifications_utils/test_logging.py
+++ b/tests/notifications_utils/test_logging.py
@@ -51,11 +51,16 @@ def test_base_json_formatter_contains_service_id():
     assert service_id_filter.filter(record).service_id == "no-service-id"
 
 
-def test_scrub():
-    result = logging.scrub(
-        "This is a message with 17775554324, and also 18884449323 and also 17775554324"
-    )
-    assert (
-        result
-        == "This is a message with 1XXXXX54324, and also 1XXXXX49323 and also 1XXXXX54324"
+def test_pii_filter():
+    record = builtin_logging.LogRecord(
+        name="log thing",
+        level="info",
+        pathname="path",
+        lineno=123,
+        msg="phone1: 1555555555, phone2: 1555555554, email1: fake@fake.gov, email2: fake@fake2.fake.gov",
+        exc_info=None,
+        args=None,
     )
+    pii_filter = logging.PIIFilter()
+    clean_msg = "phone1: 1XXXXXXXXXX, phone2: 1XXXXXXXXXX, email1: XXXXX@XXXXXXX, email2: XXXXX@XXXXXXX"
+    assert pii_filter.filter(record).msg == clean_msg