[DPE-5126] Use admin server instead of the 4lw commands (#154)

Author: Batalex

poetry.lock

@@ -118,7 +118,7 @@ mccabe.max-complexity = 10
 
 [tool.pyright]
 include = ["src"]
-extraPaths = ["./lib", "src"]
+extraPaths = ["./lib"]
 pythonVersion = "3.10"
 pythonPlatform = "All"
 typeCheckingMode = "basic"

pyproject.toml

@@ -1,24 +1,24 @@
 anyio==4.4.0 ; python_version >= "3.10" and python_version < "4.0"
-boto3-stubs[s3]==1.35.15 ; python_version >= "3.10" and python_version < "4.0"
-boto3==1.35.15 ; python_version >= "3.10" and python_version < "4.0"
-botocore-stubs==1.35.15 ; python_version >= "3.10" and python_version < "4.0"
-botocore==1.35.15 ; python_version >= "3.10" and python_version < "4.0"
+boto3-stubs[s3]==1.35.19 ; python_version >= "3.10" and python_version < "4.0"
+boto3==1.35.19 ; python_version >= "3.10" and python_version < "4.0"
+botocore-stubs==1.35.19 ; python_version >= "3.10" and python_version < "4.0"
+botocore==1.35.19 ; python_version >= "3.10" and python_version < "4.0"
 certifi==2024.8.30 ; python_version >= "3.10" and python_version < "4.0"
 cffi==1.17.1 ; python_version >= "3.10" and python_version < "4.0" and platform_python_implementation != "PyPy"
-cosl==0.0.32 ; python_version >= "3.10" and python_version < "4.0"
+cosl==0.0.33 ; python_version >= "3.10" and python_version < "4.0"
 cryptography==43.0.1 ; python_version >= "3.10" and python_version < "4.0"
 exceptiongroup==1.2.2 ; python_version >= "3.10" and python_version < "3.11"
 h11==0.14.0 ; python_version >= "3.10" and python_version < "4.0"
 httpcore==1.0.5 ; python_version >= "3.10" and python_version < "4.0"
 httpx==0.27.2 ; python_version >= "3.10" and python_version < "4.0"
-idna==3.8 ; python_version >= "3.10" and python_version < "4.0"
+idna==3.10 ; python_version >= "3.10" and python_version < "4.0"
 jmespath==1.0.1 ; python_version >= "3.10" and python_version < "4.0"
 kazoo==2.9.0 ; python_version >= "3.10" and python_version < "4.0"
-lightkube-models==1.30.0.8 ; python_version >= "3.10" and python_version < "4.0"
+lightkube-models==1.31.1.8 ; python_version >= "3.10" and python_version < "4.0"
 lightkube==0.15.4 ; python_version >= "3.10" and python_version < "4.0"
 markdown-it-py==3.0.0 ; python_version >= "3.10" and python_version < "4.0"
 mdurl==0.1.2 ; python_version >= "3.10" and python_version < "4.0"
-mypy-boto3-s3==1.35.2 ; python_version >= "3.10" and python_version < "4.0"
+mypy-boto3-s3==1.35.16 ; python_version >= "3.10" and python_version < "4.0"
 ops==2.16.1 ; python_version >= "3.10" and python_version < "4.0"
 pure-sasl==0.6.2 ; python_version >= "3.10" and python_version < "4.0"
 pycparser==2.22 ; python_version >= "3.10" and python_version < "4.0" and platform_python_implementation != "PyPy"
@@ -36,5 +36,5 @@ tenacity==9.0.0 ; python_version >= "3.10" and python_version < "4.0"
 types-awscrt==0.21.5 ; python_version >= "3.10" and python_version < "4.0"
 types-s3transfer==0.10.2 ; python_version >= "3.10" and python_version < "4.0"
 typing-extensions==4.12.2 ; python_version >= "3.10" and python_version < "4.0"
-urllib3==2.2.2 ; python_version >= "3.10" and python_version < "4.0"
+urllib3==2.2.3 ; python_version >= "3.10" and python_version < "4.0"
 websocket-client==1.8.0 ; python_version >= "3.10" and python_version < "4.0"

requirements.txt

@@ -5,20 +5,18 @@
 """Implementation of WorkloadBase for running on VMs."""
 import logging
 import os
-import re
 import secrets
 import shutil
 import string
 import subprocess
-from subprocess import CalledProcessError
 
+import httpx
 from charms.operator_libs_linux.v1 import snap
-from ops.pebble import ExecError
 from tenacity import retry, retry_if_result, stop_after_attempt, wait_fixed
 from typing_extensions import override
 
 from core.workload import WorkloadBase
-from literals import CHARMED_ZOOKEEPER_SNAP_REVISION, CLIENT_PORT
+from literals import ADMIN_SERVER_PORT, CHARMED_ZOOKEEPER_SNAP_REVISION
 
 logger = logging.getLogger(__name__)
 
@@ -109,27 +107,14 @@ def healthy(self) -> bool:
         if not self.alive:
             return False
 
-        # netcat isn't a default utility, so can't guarantee it's on the charm containers
-        # this ugly hack avoids needing netcat
-        bash_netcat = (
-            f"echo '4lw' | (exec 3<>/dev/tcp/localhost/{CLIENT_PORT}; cat >&3; cat <&3; exec 3<&-)"
-        )
-        ruok = [bash_netcat.replace("4lw", "ruok")]
-        srvr = [bash_netcat.replace("4lw", "srvr")]
+        try:
+            response = httpx.get(f"http://localhost:{ADMIN_SERVER_PORT}/commands/ruok", timeout=10)
+            response.raise_for_status()
 
-        # timeout needed as it can sometimes hang forever if there's a problem
-        # for example when the endpoint is unreachable
-        timeout = ["timeout", "10s", "bash", "-c"]
+        except httpx.HTTPError:
+            return False
 
-        try:
-            ruok_response = self.exec(command=timeout + ruok)
-            if not ruok_response or "imok" not in ruok_response:
-                return False
-
-            srvr_response = self.exec(command=timeout + srvr)
-            if not srvr_response or "not currently serving requests" in srvr_response:
-                return False
-        except (ExecError, CalledProcessError):
+        if response.json().get("error", None):
             return False
 
         return True
@@ -152,7 +137,7 @@ def install(self) -> bool:
             self.zookeeper.hold()
 
             return True
-        except (snap.SnapError) as e:
+        except snap.SnapError as e:
             logger.error(str(e))
             return False
 
@@ -170,21 +155,14 @@ def get_version(self) -> str:
         if not self.healthy:
             return ""
 
-        stat = [
-            "bash",
-            "-c",
-            f"echo 'stat' | (exec 3<>/dev/tcp/localhost/{CLIENT_PORT}; cat >&3; cat <&3; exec 3<&-; )",
-        ]
-
         try:
-            stat_response = self.exec(command=stat)
-            if not stat_response:
-                return ""
+            response = httpx.get(f"http://localhost:{ADMIN_SERVER_PORT}/commands/srvr", timeout=10)
+            response.raise_for_status()
 
-            matcher = re.search(r"(?P<version>\d\.\d\.\d)", stat_response)
-            version = matcher.group("version") if matcher else ""
-
-        except (ExecError, CalledProcessError):
+        except httpx.HTTPError:
             return ""
 
-        return version
+        if not (full_version := response.json().get("version", "")):
+            return full_version
+        else:
+            return full_version.split("-")[0]

src/workload.py

@@ -4,7 +4,6 @@
 
 import json
 import logging
-import re
 import subprocess
 from pathlib import Path
 from typing import Dict, Optional
@@ -14,6 +13,8 @@
 from pytest_operator.plugin import OpsTest
 from tenacity import RetryError, Retrying, retry, stop_after_attempt, wait_fixed
 
+from literals import ADMIN_SERVER_PORT
+
 logger = logging.getLogger(__name__)
 
 METADATA = yaml.safe_load(Path("./metadata.yaml").read_text())
@@ -50,28 +51,26 @@ async def wait_idle(ops_test, apps: list[str] = [APP_NAME], units: int = 3) -> N
     stop=stop_after_attempt(60),
     reraise=True,
 )
-def srvr(host: str) -> dict:
-    """Calls srvr 4lw command to specified host.
+def srvr(model_full_name: str, unit: str) -> dict:
+    """Calls srvr 4lw command to specified unit.
 
     Args:
-        host: ZooKeeper address and port to issue srvr 4lw command to
+        model_full_name: Current test model
+        unit: ZooKeeper unit to issue srvr 4lw command to
 
     Returns:
         Dict of srvr command output key/values
     """
     response = subprocess.check_output(
-        f"echo srvr | nc {host} 2181", stderr=subprocess.PIPE, shell=True, universal_newlines=True
+        f"JUJU_MODEL={model_full_name} juju ssh {unit} sudo -i 'curl localhost:{ADMIN_SERVER_PORT}/commands/srvr -m 10'",
+        stderr=subprocess.PIPE,
+        shell=True,
+        universal_newlines=True,
     )
 
     assert response, "ZooKeeper not running"
 
-    result = {}
-    for item in response.splitlines():
-        k = re.split(": ", item)[0]
-        v = re.split(": ", item)[1]
-        result[k] = v
-
-    return result
+    return json.loads(response)
 
 
 def get_hosts_from_status(
@@ -176,13 +175,17 @@ def get_leader_name(ops_test: OpsTest, hosts: str, app_name: str = APP_NAME) ->
         String of unit name of the ZooKeeper quorum leader
     """
     for host in hosts.split(","):
+        unit_name = get_unit_name_from_host(ops_test, host, app_name)
         try:
-            mode = srvr(host.split(":")[0])["Mode"]
+            mode = (
+                srvr(ops_test.model_full_name, unit_name)
+                .get("server_stats", {})
+                .get("server_state", "")
+            )
         except subprocess.CalledProcessError:  # unit is down
             continue
         if mode == "leader":
-            leader_name = get_unit_name_from_host(ops_test, host, app_name)
-            return leader_name
+            return unit_name
 
     return ""
 
@@ -481,8 +484,12 @@ def ping_servers(ops_test: OpsTest) -> bool:
         True if all units are in quorum. Otherwise False
     """
     for unit in ops_test.model.applications[APP_NAME].units:
-        host = unit.public_address
-        mode = srvr(host)["Mode"]
+        srvr_response = srvr(ops_test.model_full_name, unit.name)
+
+        if srvr_response.get("error", None):
+            return False
+
+        mode = srvr_response.get("server_stats", {}).get("server_state", "")
         if mode not in ["leader", "follower"]:
             return False
 

tests/integration/ha/helpers.py

@@ -14,6 +14,7 @@
 from pytest_operator.plugin import OpsTest
 
 from core.workload import ZKPaths
+from literals import ADMIN_SERVER_PORT
 
 METADATA = yaml.safe_load(Path("./metadata.yaml").read_text())
 APP_NAME = METADATA["name"]
@@ -131,31 +132,32 @@ def check_key(host: str, password: str, username: str = "super") -> None:
     raise KeyError
 
 
-def srvr(host: str) -> Dict:
+def srvr(model_full_name: str, unit: str) -> dict:
     """Retrieves attributes returned from the 'srvr' 4lw command.
 
     Specifically for this test, we are interested in the "Mode" of the ZK server,
     which allows checking quorum leadership and follower active status.
     """
     response = check_output(
-        f"echo srvr | nc {host} 2181", stderr=PIPE, shell=True, universal_newlines=True
+        f"JUJU_MODEL={model_full_name} juju ssh {unit} sudo -i 'curl localhost:{ADMIN_SERVER_PORT}/commands/srvr -m 10'",
+        stderr=PIPE,
+        shell=True,
+        universal_newlines=True,
     )
 
     assert response, "ZooKeeper not running"
 
-    result = {}
-    for item in response.splitlines():
-        k = re.split(": ", item)[0]
-        v = re.split(": ", item)[1]
-        result[k] = v
-
-    return result
+    return json.loads(response)
 
 
 async def ping_servers(ops_test: OpsTest) -> bool:
     for unit in ops_test.model.applications[APP_NAME].units:
-        host = unit.public_address
-        mode = srvr(host)["Mode"]
+        srvr_response = srvr(ops_test.model_full_name, unit.name)
+
+        if srvr_response.get("error", None):
+            return False
+
+        mode = srvr_response.get("server_stats", {}).get("server_state", "")
         if mode not in ["leader", "follower"]:
             return False
 
@@ -164,8 +166,9 @@ async def ping_servers(ops_test: OpsTest) -> bool:
 
 async def correct_version_running(ops_test: OpsTest, expected_version: str) -> bool:
     for unit in ops_test.model.applications[APP_NAME].units:
-        host = unit.public_address
-        if expected_version not in srvr(host)["Zookeeper version"]:
+        srvr_response = srvr(ops_test.model_full_name, unit.name)
+
+        if expected_version not in srvr_response.get("version", ""):
             return False
 
     return True

tests/integration/helpers.py

@@ -8,6 +8,7 @@
 from pathlib import Path
 from unittest.mock import DEFAULT, Mock, PropertyMock, patch
 
+import httpx
 import pytest
 import yaml
 from charms.rolling_ops.v0.rollingops import WaitingStatus
@@ -1055,16 +1056,18 @@ def test_update_relation_data(harness):
 
 
 def test_workload_version_is_setted(harness, monkeypatch):
-    output_install = (
-        "Zookeeper version: 3.8.1-ubuntu0-${mvngit.commit.id}, built on 2023-11-21 15:33 UTC"
-    )
-    output_changed = (
-        "Zookeeper version: 3.8.2-ubuntu0-${mvngit.commit.id}, built on 2023-11-21 15:33 UTC"
-    )
+    output_install = {
+        "version": "3.8.1-ubuntu0-${mvngit.commit.id}, built on 2023-11-21 15:33 UTC"
+    }
+    output_changed = {
+        "version": "3.8.2-ubuntu0-${mvngit.commit.id}, built on 2023-11-21 15:33 UTC"
+    }
+    response_mock = Mock()
+    response_mock.return_value.json.side_effect = [output_install, output_changed]
     monkeypatch.setattr(
-        harness.charm.workload,
-        "exec",
-        Mock(side_effect=[output_install, output_changed]),
+        httpx,
+        "get",
+        response_mock,
     )
     monkeypatch.setattr(harness.charm.workload, "install", Mock(return_value=True))
     monkeypatch.setattr(harness.charm.workload, "healthy", Mock(return_value=True))