Merge branch 'main' into dk/1016-nameservers-ui

.github/ISSUE_TEMPLATE/issue-default.yml

@@ -12,7 +12,7 @@ body:
     attributes:
       label: Issue description
       description: |
-        Describe the issue so that someone who wasn't present for its discovery can understand why it matters. Use full sentences, plain language, and good [formatting](https://docs.github.com/en/get-started/writing-on-github/getting-started-with-writing-and-formatting-on-github/basic-writing-and-formatting-syntax).
+        Describe the issue so that someone who wasn't present for its discovery can understand why it matters. Use full sentences, plain language, and [good formatting](https://docs.github.com/en/get-started/writing-on-github/getting-started-with-writing-and-formatting-on-github/basic-writing-and-formatting-syntax).
     validations:
       required: true
   - type: textarea
@@ -31,7 +31,7 @@ body:
     attributes:
       label: Links to other issues
       description: |
-        "Add issue #numbers this relates to and how (e.g., 🚧 :construction: Blocks, ⛔️ :no_entry: Is blocked by, 🔄 :repeat: Relates to)."
+        "Add issue #numbers this relates to and how (e.g., 🚧 [construction] Blocks, ⛔️ [no_entry] Is blocked by, 🔄 [arrows_counterclockwise] Relates to)."
       placeholder: 🔄 Relates to... 
   - type: markdown
     id: note

docs/architecture/decisions/0023-use-geventconnpool.md

@@ -4,7 +4,7 @@ Date: 2023-13-10
 
 ## Status
 
-In Review
+Accepted
 
 ## Context
 

src/Pipfile

@@ -25,7 +25,10 @@ django-phonenumber-field = {extras = ["phonenumberslite"], version = "*"}
 boto3 = "*"
 typing-extensions ='*'
 django-login-required-middleware = "*"
+greenlet = "*"
+gevent = "*"
 fred-epplib = {git = "https://github.com/cisagov/epplib.git", ref = "master"}
+geventconnpool = {git = "https://github.com/rasky/geventconnpool.git", ref = "1bbb93a714a331a069adf27265fe582d9ba7ecd4"}
 
 [dev-packages]
 django-debug-toolbar = "*"

src/epplibwrapper/client.py

@@ -1,7 +1,10 @@
 """Provide a wrapper around epplib to handle authentication and errors."""
 
 import logging
+
 from time import sleep
+from gevent import Timeout
+from epplibwrapper.utility.pool_status import PoolStatus
 
 try:
     from epplib.client import Client
@@ -16,6 +19,7 @@ from django.conf import settings
 from .cert import Cert, Key
 from .errors import LoginError, RegistryError
 from .socket import Socket
+from .utility.pool import EPPConnectionPool
 
 logger = logging.getLogger(__name__)
 
@@ -39,9 +43,8 @@ class EPPLibWrapper:
     ATTN: This should not be used directly. Use `Domain` from domain.py.
     """
 
-    def __init__(self) -> None:
+    def __init__(self, start_connection_pool=True) -> None:
         """Initialize settings which will be used for all connections."""
-
         # prepare (but do not send) a Login command
         self._login = commands.Login(
             cl_id=settings.SECRET_REGISTRY_CL_ID,
@@ -51,6 +54,7 @@ class EPPLibWrapper:
                 "urn:ietf:params:xml:ns:contact-1.0",
             ],
         )
+
         # establish a client object with a TCP socket transport
         self._client = Client(
             SocketTransport(
@@ -60,37 +64,77 @@ class EPPLibWrapper:
                 password=settings.SECRET_REGISTRY_KEY_PASSPHRASE,
             )
         )
-        # prepare a context manager which will connect and login when invoked
-        # (it will also logout and disconnect when the context manager exits)
-        self._connect = Socket(self._client, self._login)
+
+        self.pool_options = {
+            # Pool size
+            "size": settings.EPP_CONNECTION_POOL_SIZE,
+            # Which errors the pool should look out for.
+            # Avoid changing this unless necessary,
+            # it can and will break things.
+            "exc_classes": (TransportError,),
+            # Occasionally pings the registry to keep the connection alive.
+            # Value in seconds => (keepalive / size)
+            "keepalive": settings.POOL_KEEP_ALIVE,
+        }
+
+        self._pool = None
+
+        # Tracks the status of the pool
+        self.pool_status = PoolStatus()
+
+        if start_connection_pool:
+            self.start_connection_pool()
 
     def _send(self, command):
         """Helper function used by `send`."""
-        try:
         cmd_type = command.__class__.__name__
-            with self._connect as wire:
-                response = wire.send(command)
+
+        # Start a timeout to check if the pool is hanging
+        timeout = Timeout(settings.POOL_TIMEOUT)
+        timeout.start()
+
+        try:
+            if not self.pool_status.connection_success:
+                raise LoginError(
+                    "Couldn't connect to the registry after three attempts"
+                )
+            with self._pool.get() as connection:
+                response = connection.send(command)
+        except Timeout as t:
+            # If more than one pool exists,
+            # multiple timeouts can be floating around.
+            # We need to be specific as to which we are targeting.
+            if t is timeout:
+                # Flag that the pool is frozen,
+                # then restart the pool.
+                self.pool_status.pool_hanging = True
+                self.start_connection_pool()
         except (ValueError, ParsingError) as err:
-            message = "%s failed to execute due to some syntax error."
-            logger.warning(message, cmd_type, exc_info=True)
+            message = f"{cmd_type} failed to execute due to some syntax error."
+            logger.error(f"{message} Error: {err}", exc_info=True)
             raise RegistryError(message) from err
         except TransportError as err:
-            message = "%s failed to execute due to a connection error."
-            logger.warning(message, cmd_type, exc_info=True)
+            message = f"{cmd_type} failed to execute due to a connection error."
+            logger.error(f"{message} Error: {err}", exc_info=True)
             raise RegistryError(message) from err
         except LoginError as err:
-            message = "%s failed to execute due to a registry login error."
-            logger.warning(message, cmd_type, exc_info=True)
+            # For linter due to it not liking this line length
+            text = "failed to execute due to a registry login error."
+            message = f"{cmd_type} {text}"
+            logger.error(f"{message} Error: {err}", exc_info=True)
             raise RegistryError(message) from err
         except Exception as err:
-            message = "%s failed to execute due to an unknown error." % err
-            logger.warning(message, cmd_type, exc_info=True)
+            message = f"{cmd_type} failed to execute due to an unknown error."
+            logger.error(f"{message} Error: {err}", exc_info=True)
             raise RegistryError(message) from err
         else:
             if response.code >= 2000:
                 raise RegistryError(response.msg, code=response.code)
             else:
                 return response
+        finally:
+            # Close the timeout no matter what happens
+            timeout.close()
 
     def send(self, command, *, cleaned=False):
         """Login, send the command, then close the connection. Tries 3 times."""
@@ -98,6 +142,23 @@ class EPPLibWrapper:
         if not cleaned:
             raise ValueError("Please sanitize user input before sending it.")
 
+        # Reopen the pool if its closed
+        # Only occurs when a login error is raised, after connection is successful
+        if not self.pool_status.pool_running:
+            # We want to reopen the connection pool,
+            # but we don't want the end user to wait while it opens.
+            # Raise syntax doesn't allow this, so we use a try/catch
+            # block.
+            try:
+                logger.error("Can't contact the Registry. Pool was not running.")
+                raise RegistryError("Can't contact the Registry. Pool was not running.")
+            except RegistryError as err:
+                raise err
+            finally:
+                # Code execution will halt after here.
+                # The end user will need to recall .send.
+                self.start_connection_pool()
+
         counter = 0  # we'll try 3 times
         while True:
             try:
@@ -109,11 +170,73 @@ class EPPLibWrapper:
                 else:  # don't try again
                     raise err
 
+    def get_pool(self):
+        """Get the current pool instance"""
+        return self._pool
+
+    def _create_pool(self, client, login, options):
+        """Creates and returns new pool instance"""
+        return EPPConnectionPool(client, login, options)
+
+    def start_connection_pool(self, restart_pool_if_exists=True):
+        """Starts a connection pool for the registry.
+
+        restart_pool_if_exists -> bool:
+        If an instance of the pool already exists,
+        then then that instance will be killed first.
+        It is generally recommended to keep this enabled.
+        """
+        # Since we reuse the same creds for each pool, we can test on
+        # one socket, and if successful, then we know we can connect.
+        if not self._test_registry_connection_success():
+            logger.warning("Cannot contact the Registry")
+            self.pool_status.connection_success = False
+        else:
+            self.pool_status.connection_success = True
+
+            # If this function is reinvoked, then ensure
+            # that we don't have duplicate data sitting around.
+            if self._pool is not None and restart_pool_if_exists:
+                logger.info("Connection pool restarting...")
+                self.kill_pool()
+
+            self._pool = self._create_pool(self._client, self._login, self.pool_options)
+
+            self.pool_status.pool_running = True
+            self.pool_status.pool_hanging = False
+
+            logger.info("Connection pool started")
+
+    def kill_pool(self):
+        """Kills the existing pool. Use this instead
+        of self._pool = None, as that doesn't clear
+        gevent instances."""
+        if self._pool is not None:
+            self._pool.kill_all_connections()
+            self._pool = None
+            self.pool_status.pool_running = False
+            return None
+        logger.info("kill_pool() was invoked but there was no pool to delete")
+
+    def _test_registry_connection_success(self):
+        """Check that determines if our login
+        credentials are valid, and/or if the Registrar
+        can be contacted
+        """
+        socket = Socket(self._client, self._login)
+        can_login = False
+
+        # Something went wrong if this doesn't exist
+        if hasattr(socket, "test_connection_success"):
+            can_login = socket.test_connection_success()
+
+        return can_login
+
 
 try:
     # Initialize epplib
     CLIENT = EPPLibWrapper()
-    logger.debug("registry client initialized")
+    logger.info("registry client initialized")
 except Exception:
     CLIENT = None  # type: ignore
     logger.warning(

src/epplibwrapper/socket.py

@@ -1,7 +1,9 @@
 import logging
+from time import sleep
 
 try:
     from epplib import commands
+    from epplib.client import Client
 except ImportError:
     pass
 
@@ -14,24 +16,84 @@ logger = logging.getLogger(__name__)
 class Socket:
     """Context manager which establishes a TCP connection with registry."""
 
-    def __init__(self, client, login) -> None:
+    def __init__(self, client: Client, login: commands.Login) -> None:
         """Save the epplib client and login details."""
         self.client = client
         self.login = login
 
     def __enter__(self):
+        """Runs connect(), which opens a connection with EPPLib."""
+        self.connect()
+
+    def __exit__(self, *args, **kwargs):
+        """Runs disconnect(), which closes a connection with EPPLib."""
+        self.disconnect()
+
+    def connect(self):
         """Use epplib to connect."""
         self.client.connect()
         response = self.client.send(self.login)
-        if response.code >= 2000:
+        if self.is_login_error(response.code):
             self.client.close()
             raise LoginError(response.msg)
         return self.client
 
-    def __exit__(self, *args, **kwargs):
+    def disconnect(self):
         """Close the connection."""
         try:
             self.client.send(commands.Logout())
             self.client.close()
         except Exception:
             logger.warning("Connection to registry was not cleanly closed.")
+
+    def send(self, command):
+        """Sends a command to the registry.
+        If the RegistryError code is >= 2000,
+        then this function raises a LoginError.
+        The calling function should handle this."""
+        response = self.client.send(command)
+        if self.is_login_error(response.code):
+            self.client.close()
+            raise LoginError(response.msg)
+
+        return response
+
+    def is_login_error(self, code):
+        """Returns the result of code >= 2000 for RegistryError.
+        This indicates that something weird happened on the Registry,
+        and that we should return a LoginError."""
+        return code >= 2000
+
+    def test_connection_success(self):
+        """Tests if a successful connection can be made with the registry.
+        Tries 3 times."""
+        # Something went wrong if this doesn't exist
+        if not hasattr(self.client, "connect"):
+            logger.warning("self.client does not have a connect attribute")
+            return False
+
+        counter = 0  # we'll try 3 times
+        while True:
+            try:
+                self.client.connect()
+                response = self.client.send(self.login)
+            except LoginError as err:
+                if err.should_retry() and counter < 3:
+                    counter += 1
+                    sleep((counter * 50) / 1000)  # sleep 50 ms to 150 ms
+                else:  # don't try again
+                    return False
+            # Occurs when an invalid creds are passed in - such as on localhost
+            except OSError as err:
+                logger.error(err)
+                return False
+            else:
+                self.disconnect()
+
+                # If we encounter a login error, fail
+                if self.is_login_error(response.code):
+                    logger.warning("A login error was found in test_connection_success")
+                    return False
+
+                # Otherwise, just return true
+                return True

src/epplibwrapper/tests/test_pool.py

@@ -0,0 +1,258 @@
+import datetime
+from pathlib import Path
+from unittest.mock import MagicMock, patch
+from dateutil.tz import tzlocal  # type: ignore
+from django.test import TestCase
+from epplibwrapper.client import EPPLibWrapper
+from epplibwrapper.errors import RegistryError
+from epplibwrapper.socket import Socket
+from epplibwrapper.utility.pool import EPPConnectionPool
+from registrar.models.domain import registry
+from contextlib import ExitStack
+
+import logging
+
+try:
+    from epplib import commands
+    from epplib.client import Client
+    from epplib.exceptions import TransportError
+    from epplib.transport import SocketTransport
+    from epplib.models import common, info
+except ImportError:
+    pass
+
+logger = logging.getLogger(__name__)
+
+
+class TestConnectionPool(TestCase):
+    """Tests for our connection pooling behaviour"""
+
+    def setUp(self):
+        # Mimic the settings added to settings.py
+        self.pool_options = {
+            # Current pool size
+            "size": 1,
+            # Which errors the pool should look out for
+            "exc_classes": (TransportError,),
+            # Occasionally pings the registry to keep the connection alive.
+            # Value in seconds => (keepalive / size)
+            "keepalive": 60,
+        }
+
+    def fake_socket(self, login, client):
+        # Linter reasons
+        pw = "none"
+        # Create a fake client object
+        fake_client = Client(
+            SocketTransport(
+                "none",
+                cert_file="path/to/cert_file",
+                key_file="path/to/key_file",
+                password=pw,
+            )
+        )
+
+        return Socket(fake_client, MagicMock())
+
+    def patch_success(self):
+        return True
+
+    def fake_send(self, command, cleaned=None):
+        mock = MagicMock(
+            code=1000,
+            msg="Command completed successfully",
+            res_data=None,
+            cl_tr_id="xkw1uo#2023-10-17T15:29:09.559376",
+            sv_tr_id="5CcH4gxISuGkq8eqvr1UyQ==-35a",
+            extensions=[],
+            msg_q=None,
+        )
+        return mock
+
+    def fake_client(mock_client):
+        pw = "none"
+        client = Client(
+            SocketTransport(
+                "none",
+                cert_file="path/to/cert_file",
+                key_file="path/to/key_file",
+                password=pw,
+            )
+        )
+        return client
+
+    @patch.object(EPPLibWrapper, "_test_registry_connection_success", patch_success)
+    def test_pool_sends_data(self):
+        """A .send is invoked on the pool successfully"""
+        expected_result = {
+            "cl_tr_id": None,
+            "code": 1000,
+            "extensions": [],
+            "msg": "Command completed successfully",
+            "msg_q": None,
+            "res_data": [
+                info.InfoDomainResultData(
+                    roid="DF1340360-GOV",
+                    statuses=[
+                        common.Status(
+                            state="serverTransferProhibited",
+                            description=None,
+                            lang="en",
+                        ),
+                        common.Status(state="inactive", description=None, lang="en"),
+                    ],
+                    cl_id="gov2023-ote",
+                    cr_id="gov2023-ote",
+                    cr_date=datetime.datetime(
+                        2023, 8, 15, 23, 56, 36, tzinfo=tzlocal()
+                    ),
+                    up_id="gov2023-ote",
+                    up_date=datetime.datetime(2023, 8, 17, 2, 3, 19, tzinfo=tzlocal()),
+                    tr_date=None,
+                    name="test3.gov",
+                    registrant="TuaWnx9hnm84GCSU",
+                    admins=[],
+                    nsset=None,
+                    keyset=None,
+                    ex_date=datetime.date(2024, 8, 15),
+                    auth_info=info.DomainAuthInfo(pw="2fooBAR123fooBaz"),
+                )
+            ],
+            "sv_tr_id": "wRRNVhKhQW2m6wsUHbo/lA==-29a",
+        }
+
+        # Mock a response from EPP
+        def fake_receive(command, cleaned=None):
+            location = Path(__file__).parent / "utility" / "infoDomain.xml"
+            xml = (location).read_bytes()
+            return xml
+
+        # Mock what happens inside the "with"
+        with ExitStack() as stack:
+            stack.enter_context(
+                patch.object(EPPConnectionPool, "_create_socket", self.fake_socket)
+            )
+            stack.enter_context(patch.object(Socket, "connect", self.fake_client))
+            stack.enter_context(patch.object(SocketTransport, "send", self.fake_send))
+            stack.enter_context(patch.object(SocketTransport, "receive", fake_receive))
+            # Restart the connection pool
+            registry.start_connection_pool()
+            # Pool should be running, and be the right size
+            self.assertEqual(registry.pool_status.connection_success, True)
+            self.assertEqual(registry.pool_status.pool_running, True)
+
+            # Send a command
+            result = registry.send(commands.InfoDomain(name="test.gov"), cleaned=True)
+
+            # Should this ever fail, it either means that the schema has changed,
+            # or the pool is broken.
+            # If the schema has changed: Update the associated infoDomain.xml file
+            self.assertEqual(result.__dict__, expected_result)
+
+            # The number of open pools should match the number of requested ones.
+            # If it is 0, then they failed to open
+            self.assertEqual(len(registry._pool.conn), self.pool_options["size"])
+
+    @patch.object(EPPLibWrapper, "_test_registry_connection_success", patch_success)
+    def test_pool_restarts_on_send(self):
+        """A .send is invoked, but the pool isn't running.
+        The pool should restart."""
+        expected_result = {
+            "cl_tr_id": None,
+            "code": 1000,
+            "extensions": [],
+            "msg": "Command completed successfully",
+            "msg_q": None,
+            "res_data": [
+                info.InfoDomainResultData(
+                    roid="DF1340360-GOV",
+                    statuses=[
+                        common.Status(
+                            state="serverTransferProhibited",
+                            description=None,
+                            lang="en",
+                        ),
+                        common.Status(state="inactive", description=None, lang="en"),
+                    ],
+                    cl_id="gov2023-ote",
+                    cr_id="gov2023-ote",
+                    cr_date=datetime.datetime(
+                        2023, 8, 15, 23, 56, 36, tzinfo=tzlocal()
+                    ),
+                    up_id="gov2023-ote",
+                    up_date=datetime.datetime(2023, 8, 17, 2, 3, 19, tzinfo=tzlocal()),
+                    tr_date=None,
+                    name="test3.gov",
+                    registrant="TuaWnx9hnm84GCSU",
+                    admins=[],
+                    nsset=None,
+                    keyset=None,
+                    ex_date=datetime.date(2024, 8, 15),
+                    auth_info=info.DomainAuthInfo(pw="2fooBAR123fooBaz"),
+                )
+            ],
+            "sv_tr_id": "wRRNVhKhQW2m6wsUHbo/lA==-29a",
+        }
+
+        # Mock a response from EPP
+        def fake_receive(command, cleaned=None):
+            location = Path(__file__).parent / "utility" / "infoDomain.xml"
+            xml = (location).read_bytes()
+            return xml
+
+        # Mock what happens inside the "with"
+        with ExitStack() as stack:
+            stack.enter_context(
+                patch.object(EPPConnectionPool, "_create_socket", self.fake_socket)
+            )
+            stack.enter_context(patch.object(Socket, "connect", self.fake_client))
+            stack.enter_context(patch.object(SocketTransport, "send", self.fake_send))
+            stack.enter_context(patch.object(SocketTransport, "receive", fake_receive))
+            # Kill the connection pool
+            registry.kill_pool()
+
+            self.assertEqual(registry.pool_status.connection_success, False)
+            self.assertEqual(registry.pool_status.pool_running, False)
+
+            # An exception should be raised as end user will be informed
+            # that they cannot connect to EPP
+            with self.assertRaises(RegistryError):
+                expected = "InfoDomain failed to execute due to a connection error."
+                result = registry.send(
+                    commands.InfoDomain(name="test.gov"), cleaned=True
+                )
+                self.assertEqual(result, expected)
+
+            # A subsequent command should be successful, as the pool restarts
+            result = registry.send(commands.InfoDomain(name="test.gov"), cleaned=True)
+            # Should this ever fail, it either means that the schema has changed,
+            # or the pool is broken.
+            # If the schema has changed: Update the associated infoDomain.xml file
+            self.assertEqual(result.__dict__, expected_result)
+
+            # The number of open pools should match the number of requested ones.
+            # If it is 0, then they failed to open
+            self.assertEqual(len(registry._pool.conn), self.pool_options["size"])
+
+    @patch.object(EPPLibWrapper, "_test_registry_connection_success", patch_success)
+    def test_raises_connection_error(self):
+        """A .send is invoked on the pool, but registry connection is lost
+        right as we send a command."""
+
+        with ExitStack() as stack:
+            stack.enter_context(
+                patch.object(EPPConnectionPool, "_create_socket", self.fake_socket)
+            )
+            stack.enter_context(patch.object(Socket, "connect", self.fake_client))
+
+            # Pool should be running
+            self.assertEqual(registry.pool_status.connection_success, True)
+            self.assertEqual(registry.pool_status.pool_running, True)
+
+            # Try to send a command out - should fail
+            with self.assertRaises(RegistryError):
+                expected = "InfoDomain failed to execute due to a connection error."
+                result = registry.send(
+                    commands.InfoDomain(name="test.gov"), cleaned=True
+                )
+                self.assertEqual(result, expected)

src/epplibwrapper/tests/utility/infoDomain.xml

@@ -0,0 +1,33 @@
+<?xml version="1.0" encoding="UTF-8" standalone="yes"?>
+
+<epp xmlns:domain="urn:ietf:params:xml:ns:domain-1.0" xmlns:contact="urn:ietf:params:xml:ns:contact-1.0" xmlns:fee="urn:ietf:params:xml:ns:fee-0.6" xmlns:packageToken="urn:google:params:xml:ns:packageToken-1.0" xmlns="urn:ietf:params:xml:ns:epp-1.0" xmlns:fee11="urn:ietf:params:xml:ns:fee-0.11" xmlns:fee12="urn:ietf:params:xml:ns:fee-0.12" xmlns:launch="urn:ietf:params:xml:ns:launch-1.0" xmlns:secDNS="urn:ietf:params:xml:ns:secDNS-1.1" xmlns:host="urn:ietf:params:xml:ns:host-1.0">
+   <response>
+       <result code="1000">
+           <msg>Command completed successfully</msg>
+       </result>
+       <resData>
+        <domain:infData>
+            <domain:name>test3.gov</domain:name>
+            <domain:roid>DF1340360-GOV</domain:roid>
+            <domain:status s="serverTransferProhibited"/>
+            <domain:status s="inactive"/>
+            <domain:registrant>TuaWnx9hnm84GCSU</domain:registrant>
+            <domain:contact type="security">CONT2</domain:contact>
+            <domain:contact type="tech">CONT3</domain:contact>
+            <domain:clID>gov2023-ote</domain:clID>
+            <domain:crID>gov2023-ote</domain:crID>
+            <domain:crDate>2023-08-15T23:56:36Z</domain:crDate>
+            <domain:upID>gov2023-ote</domain:upID>
+            <domain:upDate>2023-08-17T02:03:19Z</domain:upDate>
+            <domain:exDate>2024-08-15T23:56:36Z</domain:exDate>
+            <domain:authInfo>
+                <domain:pw>2fooBAR123fooBaz</domain:pw>
+            </domain:authInfo>
+        </domain:infData>
+    </resData>
+       <trID>
+           <svTRID>wRRNVhKhQW2m6wsUHbo/lA==-29a</svTRID>
+       </trID>
+   </response>
+
+</epp>

src/epplibwrapper/utility/pool.py

@@ -0,0 +1,134 @@
+import logging
+from typing import List
+import gevent
+from geventconnpool import ConnectionPool
+from epplibwrapper.socket import Socket
+from epplibwrapper.utility.pool_error import PoolError, PoolErrorCodes
+
+try:
+    from epplib.commands import Hello
+    from epplib.exceptions import TransportError
+except ImportError:
+    pass
+
+from gevent.lock import BoundedSemaphore
+from collections import deque
+
+logger = logging.getLogger(__name__)
+
+
+class EPPConnectionPool(ConnectionPool):
+    """A connection pool for EPPLib.
+
+    Args:
+        client (Client): The client
+        login (commands.Login): Login creds
+        options (dict): Options for the ConnectionPool
+        base class
+    """
+
+    def __init__(self, client, login, options: dict):
+        # For storing shared credentials
+        self._client = client
+        self._login = login
+
+        # Keep track of each greenlet
+        self.greenlets: List[gevent.Greenlet] = []
+
+        # Define optional pool settings.
+        # Kept in a dict so that the parent class,
+        # client.py, can maintain seperation/expandability
+        self.size = 1
+        if "size" in options:
+            self.size = options["size"]
+
+        self.exc_classes = tuple((TransportError,))
+        if "exc_classes" in options:
+            self.exc_classes = options["exc_classes"]
+
+        self.keepalive = None
+        if "keepalive" in options:
+            self.keepalive = options["keepalive"]
+
+        # Determines the period in which new
+        # gevent threads are spun up.
+        # This time period is in seconds. So for instance, .1 would be .1 seconds.
+        self.spawn_frequency = 0.1
+        if "spawn_frequency" in options:
+            self.spawn_frequency = options["spawn_frequency"]
+
+        self.conn: deque = deque()
+        self.lock = BoundedSemaphore(self.size)
+
+        self.populate_all_connections()
+
+    def _new_connection(self):
+        socket = self._create_socket(self._client, self._login)
+        try:
+            connection = socket.connect()
+            return connection
+        except Exception as err:
+            message = f"Failed to execute due to a registry error: {err}"
+            logger.error(message, exc_info=True)
+            # We want to raise a pool error rather than a LoginError here
+            # because if this occurs internally, we should handle this
+            # differently than we otherwise would for LoginError.
+            raise PoolError(code=PoolErrorCodes.NEW_CONNECTION_FAILED) from err
+
+    def _keepalive(self, c):
+        """Sends a command to the server to keep the connection alive."""
+        try:
+            # Sends a ping to the registry via EPPLib
+            c.send(Hello())
+        except Exception as err:
+            message = "Failed to keep the connection alive."
+            logger.error(message, exc_info=True)
+            raise PoolError(code=PoolErrorCodes.KEEP_ALIVE_FAILED) from err
+
+    def _create_socket(self, client, login) -> Socket:
+        """Creates and returns a socket instance"""
+        socket = Socket(client, login)
+        return socket
+
+    def get_connections(self):
+        """Returns the connection queue"""
+        return self.conn
+
+    def kill_all_connections(self):
+        """Kills all active connections in the pool."""
+        try:
+            if len(self.conn) > 0:
+                gevent.killall(self.greenlets)
+
+                self.greenlets.clear()
+                self.conn.clear()
+
+                # Clear the semaphore
+                self.lock = BoundedSemaphore(self.size)
+            else:
+                logger.info("No connections to kill.")
+        except Exception as err:
+            logger.error("Could not kill all connections.")
+            raise PoolError(code=PoolErrorCodes.KILL_ALL_FAILED) from err
+
+    def populate_all_connections(self):
+        """Generates the connection pool.
+        If any connections exist, kill them first.
+        Based off of the __init__ definition for geventconnpool.
+        """
+        if len(self.conn) > 0:
+            self.kill_all_connections()
+
+        # Setup the lock
+        for i in range(self.size):
+            self.lock.acquire()
+
+        # Open multiple connections
+        for i in range(self.size):
+            self.greenlets.append(
+                gevent.spawn_later(self.spawn_frequency * i, self._addOne)
+            )
+
+        # Open a "keepalive" thread if we want to ping open connections
+        if self.keepalive:
+            self.greenlets.append(gevent.spawn(self._keepalive_periodic))

src/epplibwrapper/utility/pool_error.py

@@ -0,0 +1,52 @@
+from enum import IntEnum
+
+
+class PoolErrorCodes(IntEnum):
+    """Used in the PoolError class for
+    error mapping.
+
+    Overview of contact error codes:
+        - 2000 KILL_ALL_FAILED
+        - 2001 NEW_CONNECTION_FAILED
+        - 2002 KEEP_ALIVE_FAILED
+    """
+
+    KILL_ALL_FAILED = 2000
+    NEW_CONNECTION_FAILED = 2001
+    KEEP_ALIVE_FAILED = 2002
+
+
+class PoolError(Exception):
+    """
+    Overview of contact error codes:
+        - 2000 KILL_ALL_FAILED
+        - 2001 NEW_CONNECTION_FAILED
+        - 2002 KEEP_ALIVE_FAILED
+
+    Note: These are separate from the error codes returned from EppLib
+    """
+
+    # Used variables due to linter requirements
+    kill_failed = "Could not kill all connections. Are multiple pools running?"
+    conn_failed = (
+        "Failed to execute due to a registry error."
+        " See previous logs to determine the cause of the error."
+    )
+    alive_failed = (
+        "Failed to keep the connection alive. "
+        "It is likely that the registry returned a LoginError."
+    )
+    _error_mapping = {
+        PoolErrorCodes.KILL_ALL_FAILED: kill_failed,
+        PoolErrorCodes.NEW_CONNECTION_FAILED: conn_failed,
+        PoolErrorCodes.KEEP_ALIVE_FAILED: alive_failed,
+    }
+
+    def __init__(self, *args, code=None, **kwargs):
+        super().__init__(*args, **kwargs)
+        self.code = code
+        if self.code in self._error_mapping:
+            self.message = self._error_mapping.get(self.code)
+
+    def __str__(self):
+        return f"{self.message}"

src/epplibwrapper/utility/pool_status.py

@@ -0,0 +1,12 @@
+class PoolStatus:
+    """A list of Booleans to keep track of Pool Status.
+
+    pool_running -> bool: Tracks if the pool itself is active or not.
+    connection_success -> bool: Tracks if connection is possible with the registry.
+    pool_hanging -> pool: Tracks if the pool has exceeded its timeout period.
+    """
+
+    def __init__(self):
+        self.pool_running = False
+        self.connection_success = False
+        self.pool_hanging = False

src/registrar/config/settings.py

@@ -534,6 +534,20 @@ SECRET_REGISTRY_KEY = secret_registry_key
 SECRET_REGISTRY_KEY_PASSPHRASE = secret_registry_key_passphrase
 SECRET_REGISTRY_HOSTNAME = secret_registry_hostname
 
+# Use this variable to set the size of our connection pool in client.py
+# WARNING: Setting this value too high could cause frequent app crashes!
+# Having too many connections open could cause the sandbox to timeout,
+# as the spinup time could exceed the timeout time.
+EPP_CONNECTION_POOL_SIZE = 1
+
+# Determines the interval in which we ping open connections in seconds
+# Calculated as POOL_KEEP_ALIVE / EPP_CONNECTION_POOL_SIZE
+POOL_KEEP_ALIVE = 60
+
+# Determines how long we try to keep a pool alive for,
+# before restarting it.
+POOL_TIMEOUT = 60
+
 # endregion
 # region: Security and Privacy----------------------------------------------###
 

src/registrar/models/__init__.py

@@ -1,5 +1,4 @@
 from auditlog.registry import auditlog  # type: ignore
-
 from .contact import Contact
 from .domain_application import DomainApplication
 from .domain_information import DomainInformation

src/registrar/models/domain.py

@@ -9,6 +9,14 @@ from django_fsm import FSMField, transition, TransitionNotAllowed  # type: ignor
 
 from django.db import models
 from typing import Any
+
+
+from registrar.utility.errors import (
+    ActionNotAllowed,
+    NameserverError,
+    NameserverErrorCodes as nsErrorCodes,
+)
+
 from epplibwrapper import (
     CLIENT as registry,
     commands,
@@ -19,15 +27,8 @@ from epplibwrapper import (
     ErrorCode,
 )
 
-from registrar.utility.errors import (
-    ActionNotAllowed,
-    NameserverError,
-    NameserverErrorCodes as nsErrorCodes,
-)
-
 from registrar.models.utility.contact_error import ContactError, ContactErrorCodes
 
-
 from .utility.domain_field import DomainField
 from .utility.domain_helper import DomainHelper
 from .utility.time_stamped_model import TimeStampedModel

src/requirements.txt

@@ -1,53 +1,61 @@
 -i https://pypi.python.org/simple
-asgiref==3.7.2 ; python_version >= '3.7'
-boto3==1.26.145
-botocore==1.29.145 ; python_version >= '3.7'
-cachetools==5.3.1
-certifi==2023.7.22 ; python_version >= '3.6'
+annotated-types==0.6.0; python_version >= '3.8'
+asgiref==3.7.2; python_version >= '3.7'
+boto3==1.28.66; python_version >= '3.7'
+botocore==1.31.66; python_version >= '3.7'
+cachetools==5.3.1; python_version >= '3.7'
+certifi==2023.7.22; python_version >= '3.6'
 cfenv==0.5.3
-cffi==1.15.1
-charset-normalizer==3.1.0 ; python_full_version >= '3.7.0'
-cryptography==41.0.4 ; python_version >= '3.7'
-defusedxml==0.7.1 ; python_version >= '2.7' and python_version not in '3.0, 3.1, 3.2, 3.3, 3.4'
-dj-database-url==2.0.0
+cffi==1.16.0; python_version >= '3.8'
+charset-normalizer==3.3.0; python_full_version >= '3.7.0'
+cryptography==41.0.4; python_version >= '3.7'
+defusedxml==0.7.1; python_version >= '2.7' and python_version not in '3.0, 3.1, 3.2, 3.3, 3.4'
+dj-database-url==2.1.0
 dj-email-url==1.0.6
-django==4.2.3
-django-allow-cidr==0.6.0
-django-auditlog==2.3.0
+django==4.2.6; python_version >= '3.8'
+django-allow-cidr==0.7.1
+django-auditlog==2.3.0; python_version >= '3.7'
 django-cache-url==3.4.4
 django-csp==3.7
 django-fsm==2.8.1
 django-login-required-middleware==0.9.0
-django-phonenumber-field[phonenumberslite]==7.1.0
-django-widget-tweaks==1.4.12
-environs[django]==9.5.0
-faker==18.10.0
-git+https://github.com/cisagov/epplib.git@d56d183f1664f34c40ca9716a3a9a345f0ef561c#egg=fred-epplib
+django-phonenumber-field[phonenumberslite]==7.2.0; python_version >= '3.8'
+django-widget-tweaks==1.5.0; python_version >= '3.8'
+environs[django]==9.5.0; python_version >= '3.6'
+faker==19.11.0; python_version >= '3.8'
+fred-epplib@ git+https://github.com/cisagov/epplib.git@d56d183f1664f34c40ca9716a3a9a345f0ef561c
 furl==2.1.3
-future==0.18.3 ; python_version >= '2.6' and python_version not in '3.0, 3.1, 3.2, 3.3'
-gunicorn==20.1.0
-idna==3.4 ; python_version >= '3.5'
-jmespath==1.0.1 ; python_version >= '3.7'
-lxml==4.9.2 ; python_version >= '2.7' and python_version not in '3.0, 3.1, 3.2, 3.3, 3.4'
-mako==1.2.4 ; python_version >= '3.7'
-markupsafe==2.1.2 ; python_version >= '3.7'
-marshmallow==3.19.0 ; python_version >= '3.7'
-oic==1.6.0
+future==0.18.3; python_version >= '2.6' and python_version not in '3.0, 3.1, 3.2, 3.3'
+gevent==23.9.1; python_version >= '3.8'
+geventconnpool@ git+https://github.com/rasky/geventconnpool.git@1bbb93a714a331a069adf27265fe582d9ba7ecd4
+greenlet==3.0.0; python_version >= '3.7'
+gunicorn==21.2.0; python_version >= '3.5'
+idna==3.4; python_version >= '3.5'
+jmespath==1.0.1; python_version >= '3.7'
+lxml==4.9.3; python_version >= '2.7' and python_version not in '3.0, 3.1, 3.2, 3.3, 3.4'
+mako==1.2.4; python_version >= '3.7'
+markupsafe==2.1.3; python_version >= '3.7'
+marshmallow==3.20.1; python_version >= '3.8'
+oic==1.6.1; python_version ~= '3.7'
 orderedmultidict==1.0.1
-packaging==23.1 ; python_version >= '3.7'
-phonenumberslite==8.13.13
-psycopg2-binary==2.9.6
+packaging==23.2; python_version >= '3.7'
+phonenumberslite==8.13.23
+psycopg2-binary==2.9.9; python_version >= '3.7'
 pycparser==2.21
-pycryptodomex==3.18.0
-pydantic==1.10.8 ; python_version >= '3.7'
+pycryptodomex==3.19.0; python_version >= '2.7' and python_version not in '3.0, 3.1, 3.2, 3.3, 3.4'
+pydantic==2.4.2; python_version >= '3.7'
+pydantic-core==2.10.1; python_version >= '3.7'
+pydantic-settings==2.0.3; python_version >= '3.7'
 pyjwkest==1.4.2
-python-dateutil==2.8.2 ; python_version >= '2.7' and python_version not in '3.0, 3.1, 3.2, 3.3'
-python-dotenv==1.0.0 ; python_version >= '3.8'
-requests==2.31.0
-s3transfer==0.6.1 ; python_version >= '3.7'
-setuptools==67.8.0 ; python_version >= '3.7'
-six==1.16.0 ; python_version >= '2.7' and python_version not in '3.0, 3.1, 3.2, 3.3'
-sqlparse==0.4.4 ; python_version >= '3.5'
-typing-extensions==4.6.3
-urllib3==1.26.17 ; python_version >= '2.7' and python_version not in '3.0, 3.1, 3.2, 3.3, 3.4, 3.5'
-whitenoise==6.4.0
+python-dateutil==2.8.2; python_version >= '2.7' and python_version not in '3.0, 3.1, 3.2, 3.3'
+python-dotenv==1.0.0; python_version >= '3.8'
+requests==2.31.0; python_version >= '3.7'
+s3transfer==0.7.0; python_version >= '3.7'
+setuptools==68.2.2; python_version >= '3.8'
+six==1.16.0; python_version >= '2.7' and python_version not in '3.0, 3.1, 3.2, 3.3'
+sqlparse==0.4.4; python_version >= '3.5'
+typing-extensions==4.8.0; python_version >= '3.8'
+urllib3==2.0.7; python_version >= '3.7'
+whitenoise==6.6.0; python_version >= '3.8'
+zope.event==5.0; python_version >= '3.7'
+zope.interface==6.1; python_version >= '3.7'