refactor cgroup parsing to use regexes, heavily inspired by elastic/a…

…pm-agent-go#342

(as in, pretty much copied from elastic/apm-agent-go#342)

elasticapm/base.py

@@ -25,7 +25,7 @@
 from elasticapm.conf import Config, constants
 from elasticapm.conf.constants import ERROR
 from elasticapm.traces import Tracer, get_transaction
-from elasticapm.utils import compat, docker, is_master_process, stacks, varmap
+from elasticapm.utils import cgroup, compat, is_master_process, stacks, varmap
 from elasticapm.utils.encoding import keyword_field, shorten, transform
 from elasticapm.utils.module_import import import_string
 
@@ -250,7 +250,7 @@ def get_system_info(self):
             "architecture": platform.machine(),
             "platform": platform.system().lower(),
         }
-        system_data.update(docker.get_docker_metadata())
+        system_data.update(cgroup.get_cgroup_container_metadata())
         pod_name = os.environ.get("KUBERNETES_POD_NAME") or system_data["hostname"]
         changed = False
         if "kubernetes" in system_data:

elasticapm/utils/cgroup.py

@@ -0,0 +1,75 @@
+import os
+import re
+
+CGROUP_PATH = "/proc/self/cgroup"
+
+SYSTEMD_SCOPE_SUFFIX = ".scope"
+
+kubepods_regexp = re.compile(
+    r"(?:^/kubepods/[^/]+/pod([^/]+)$)|(?:^/kubepods\.slice/kubepods-[^/]+\.slice/kubepods-[^/]+-pod([^/]+)\.slice$)"
+)
+
+container_id_regexp = re.compile("^[0-9A-Fa-f]{64}$")
+
+
+def get_cgroup_container_metadata():
+    """
+    Reads docker/kubernetes metadata (container id, pod id) from /proc/self/cgroup
+
+    The result is a nested dictionary with the detected IDs, e.g.
+
+        {
+            "container": {"id": "2227daf62df6694645fee5df53c1f91271546a9560e8600a525690ae252b7f63"},
+            "pod": {"uid": "90d81341_92de_11e7_8cf2_507b9d4141fa"}
+        }
+
+    :return: a dictionary with the detected ids or {}
+    """
+    if not os.path.exists(CGROUP_PATH):
+        return {}
+    with open(CGROUP_PATH) as f:
+        return parse_cgroups(f) or {}
+
+
+def parse_cgroups(filehandle):
+    """
+    Reads lines from a file handle and tries to parse docker container IDs and kubernetes Pod IDs.
+
+    See tests.utils.docker_tests.test_cgroup_parsing for a set of test cases
+
+    :param filehandle:
+    :return: nested dictionary or None
+    """
+    for line in filehandle:
+        parts = line.strip().split(":")
+        if len(parts) != 3:
+            continue
+        cgroup_path = parts[2]
+
+        # Depending on the filesystem driver used for cgroup
+        # management, the paths in /proc/pid/cgroup will have
+        # one of the following formats in a Docker container:
+        #
+        #   systemd: /system.slice/docker-<container-ID>.scope
+        #   cgroupfs: /docker/<container-ID>
+        #
+        # In a Kubernetes pod, the cgroup path will look like:
+        #
+        #   systemd:
+        #      /kubepods.slice/kubepods-<QoS-class>.slice/kubepods-<QoS-class>-pod<pod-UID>.slice/<container-iD>.scope
+        #   cgroupfs:
+        #      /kubepods/<QoS-class>/pod<pod-UID>/<container-iD>
+
+        directory, container_id = os.path.split(cgroup_path)
+        if container_id.endswith(SYSTEMD_SCOPE_SUFFIX):
+            container_id = container_id[: -len(SYSTEMD_SCOPE_SUFFIX)]
+            if "-" in container_id:
+                container_id = container_id.split("-", 1)[1]
+        kubepods_match = kubepods_regexp.match(directory)
+        if kubepods_match:
+            pod_id = kubepods_match.group(1)
+            if not pod_id:
+                pod_id = kubepods_match.group(2)
+            return {"container": {"id": container_id}, "pod": {"uid": pod_id}}
+        elif container_id_regexp.match(container_id):
+            return {"container": {"id": container_id}}

tests/client/client_tests.py

@@ -40,15 +40,15 @@ def test_process_info(elasticapm_client):
 
 def test_system_info(elasticapm_client):
     # mock docker/kubernetes data here to get consistent behavior if test is run in docker
-    with mock.patch("elasticapm.utils.docker.get_docker_metadata") as mocked:
+    with mock.patch("elasticapm.utils.cgroup.get_cgroup_container_metadata") as mocked:
         mocked.return_value = {}
         system_info = elasticapm_client.get_system_info()
     assert {"hostname", "architecture", "platform"} == set(system_info.keys())
 
 
 def test_docker_kubernetes_system_info(elasticapm_client):
     # mock docker/kubernetes data here to get consistent behavior if test is run in docker
-    with mock.patch("elasticapm.utils.docker.get_docker_metadata") as mock_metadata, mock.patch(
+    with mock.patch("elasticapm.utils.cgroup.get_cgroup_container_metadata") as mock_metadata, mock.patch(
         "socket.gethostname"
     ) as mock_gethostname:
         mock_metadata.return_value = {"container": {"id": "123"}, "kubernetes": {"pod": {"uid": "456"}}}
@@ -71,7 +71,7 @@ def test_docker_kubernetes_system_info_from_environ():
     # initialize agent only after overriding environment
     elasticapm_client = Client()
     # mock docker/kubernetes data here to get consistent behavior if test is run in docker
-    with mock.patch("elasticapm.utils.docker.get_docker_metadata") as mock_metadata:
+    with mock.patch("elasticapm.utils.cgroup.get_cgroup_container_metadata") as mock_metadata:
         mock_metadata.return_value = {}
         system_info = elasticapm_client.get_system_info()
     assert "kubernetes" in system_info
@@ -95,7 +95,7 @@ def test_docker_kubernetes_system_info_from_environ_overrides_cgroups():
     # initialize agent only after overriding environment
     elasticapm_client = Client()
     # mock docker/kubernetes data here to get consistent behavior if test is run in docker
-    with mock.patch("elasticapm.utils.docker.get_docker_metadata") as mock_metadata, mock.patch(
+    with mock.patch("elasticapm.utils.cgroup.get_cgroup_container_metadata") as mock_metadata, mock.patch(
         "socket.gethostname"
     ) as mock_gethostname:
         mock_metadata.return_value = {"container": {"id": "123"}, "kubernetes": {"pod": {"uid": "456"}}}

tests/utils/docker_tests.py → tests/utils/cgroup_tests.py

@@ -1,7 +1,7 @@
 import mock
 import pytest
 
-from elasticapm.utils import compat, docker
+from elasticapm.utils import cgroup, compat
 
 
 @pytest.mark.parametrize(
@@ -33,5 +33,5 @@
 )
 def test_cgroup_parsing(test_input, expected):
     f = compat.StringIO(test_input)
-    result = docker.parse_cgroups(f)
-    assert expected == result
+    result = cgroup.parse_cgroups(f)
+    assert result == expected