[#1] agent模块开发

This commit is contained in:
2025-09-24 04:19:37 +00:00
parent 5e5342ce20
commit bc016826f1
25 changed files with 1123 additions and 0 deletions

View File

@@ -0,0 +1,22 @@
#!/usr/bin/env bash
set -euo pipefail
SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
SCRIPTS=(
"01_bootstrap.sh"
"02_up.sh"
"03_wait_and_assert_registration.sh"
"04_write_health_files.sh"
"05_assert_status_on_master.sh"
"06_restart_agent_and_reregister.sh"
"07_down.sh"
)
for script in "${SCRIPTS[@]}"; do
echo "[TEST] Running $script"
"$SCRIPT_DIR/$script"
echo "[TEST] $script completed"
echo
done
echo "[TEST] Agent module E2E tests completed"

View File

@@ -0,0 +1,44 @@
#!/usr/bin/env bash
set -euo pipefail
SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
TEST_ROOT="$(cd "$SCRIPT_DIR/.." && pwd)"
AGENT_ROOT="$(cd "$TEST_ROOT/.." && pwd)"
MASTER_ROOT="$(cd "$AGENT_ROOT/../master" && pwd)"
PRIVATE_ROOT="$TEST_ROOT/private"
TMP_ROOT="$TEST_ROOT/tmp"
AGENT_HOSTNAME="dev-e2euser-e2einst-pod-0"
AGENT_CONFIG_DIR="$PRIVATE_ROOT/argus/agent/$AGENT_HOSTNAME"
AGENT_HEALTH_DIR="$PRIVATE_ROOT/argus/agent/health/$AGENT_HOSTNAME"
MASTER_PRIVATE_DIR="$PRIVATE_ROOT/argus/master"
METRIC_PRIVATE_DIR="$PRIVATE_ROOT/argus/metric/prometheus"
mkdir -p "$AGENT_CONFIG_DIR"
mkdir -p "$AGENT_HEALTH_DIR"
mkdir -p "$MASTER_PRIVATE_DIR"
mkdir -p "$METRIC_PRIVATE_DIR"
mkdir -p "$TMP_ROOT"
cat > "$AGENT_CONFIG_DIR/config" <<CONFIG
HOSTNAME=$AGENT_HOSTNAME
NODE_FILE=/private/argus/agent/$AGENT_HOSTNAME/node.json
VERSION=1.1.0
MASTER_ENDPOINT=http://master:3000
REPORT_INTERVAL_SECONDS=2
SUBMODULE_HEALTH_FILE_DIR=/private/argus/agent/health/{hostname}/
GPU_NUMBER=
IP_OVERRIDE=
CONFIG
touch "$AGENT_HEALTH_DIR/.keep"
pushd "$MASTER_ROOT" >/dev/null
./scripts/build_images.sh --tag argus-master:dev
popd >/dev/null
pushd "$AGENT_ROOT" >/dev/null
./scripts/build_images.sh --tag argus-agent:dev
popd >/dev/null
echo "[INFO] Agent E2E bootstrap complete"

View File

@@ -0,0 +1,24 @@
#!/usr/bin/env bash
set -euo pipefail
SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
TEST_ROOT="$(cd "$SCRIPT_DIR/.." && pwd)"
compose() {
if docker compose version >/dev/null 2>&1; then
docker compose "$@"
else
docker-compose "$@"
fi
}
docker container rm -f argus-agent-e2e >/dev/null 2>&1 || true
docker network rm tests_default >/dev/null 2>&1 || true
pushd "$TEST_ROOT" >/dev/null
compose down --remove-orphans || true
compose up -d
popd >/dev/null
echo "[INFO] Master+Agent stack started"

View File

@@ -0,0 +1,65 @@
#!/usr/bin/env bash
set -euo pipefail
SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
TEST_ROOT="$(cd "$SCRIPT_DIR/.." && pwd)"
TMP_ROOT="$TEST_ROOT/tmp"
API_BASE="http://localhost:32300/api/v1/master"
AGENT_HOSTNAME="dev-e2euser-e2einst-pod-0"
NODE_FILE="$TEST_ROOT/private/argus/agent/$AGENT_HOSTNAME/node.json"
mkdir -p "$TMP_ROOT"
node_id=""
for _ in {1..30}; do
sleep 2
response=$(curl -sS "$API_BASE/nodes" || true)
if [[ -z "$response" ]]; then
continue
fi
list_file="$TMP_ROOT/nodes_list.json"
echo "$response" > "$list_file"
node_id=$(python3 - "$list_file" <<'PY'
import json, sys
with open(sys.argv[1]) as handle:
nodes = json.load(handle)
print(nodes[0]["id"] if nodes else "")
PY
)
if [[ -n "$node_id" ]]; then
break
fi
done
if [[ -z "$node_id" ]]; then
echo "[ERROR] Agent did not register within timeout" >&2
exit 1
fi
echo "$node_id" > "$TMP_ROOT/node_id"
if [[ ! -f "$NODE_FILE" ]]; then
echo "[ERROR] node.json not created at $NODE_FILE" >&2
exit 1
fi
python3 - "$NODE_FILE" <<'PY'
import json, sys
with open(sys.argv[1]) as handle:
node = json.load(handle)
assert "id" in node and node["id"], "node.json missing id"
PY
detail_file="$TMP_ROOT/initial_detail.json"
curl -sS "$API_BASE/nodes/$node_id" -o "$detail_file"
python3 - "$detail_file" "$TMP_ROOT/initial_ip" <<'PY'
import json, sys, pathlib
with open(sys.argv[1]) as handle:
node = json.load(handle)
ip = node["meta_data"].get("ip")
if not ip:
raise SystemExit("meta_data.ip missing")
pathlib.Path(sys.argv[2]).write_text(ip)
PY
echo "[INFO] Agent registered with node id $node_id"

View File

@@ -0,0 +1,22 @@
#!/usr/bin/env bash
set -euo pipefail
SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
TEST_ROOT="$(cd "$SCRIPT_DIR/.." && pwd)"
HEALTH_DIR="$TEST_ROOT/private/argus/agent/health/dev-e2euser-e2einst-pod-0"
cat > "$HEALTH_DIR/log-fluentbit.json" <<JSON
{
"status": "healthy",
"timestamp": "2023-10-05T12:05:00Z"
}
JSON
cat > "$HEALTH_DIR/metric-node-exporter.json" <<JSON
{
"status": "healthy",
"timestamp": "2023-10-05T12:05:00Z"
}
JSON
echo "[INFO] Health files written"

View File

@@ -0,0 +1,53 @@
#!/usr/bin/env bash
set -euo pipefail
SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
TEST_ROOT="$(cd "$SCRIPT_DIR/.." && pwd)"
TMP_ROOT="$TEST_ROOT/tmp"
API_BASE="http://localhost:32300/api/v1/master"
NODE_ID="$(cat "$TMP_ROOT/node_id")"
NODES_JSON="$TEST_ROOT/private/argus/metric/prometheus/nodes.json"
success=false
detail_file="$TMP_ROOT/agent_status_detail.json"
for _ in {1..20}; do
sleep 2
if ! curl -sS "$API_BASE/nodes/$NODE_ID" -o "$detail_file"; then
continue
fi
if python3 - "$detail_file" <<'PY'
import json, sys
with open(sys.argv[1]) as handle:
node = json.load(handle)
if node["status"] != "online":
raise SystemExit(1)
health = node.get("health", {})
if "log-fluentbit" not in health or "metric-node-exporter" not in health:
raise SystemExit(1)
PY
then
success=true
break
fi
done
if [[ "$success" != true ]]; then
echo "[ERROR] Node did not report health data in time" >&2
exit 1
fi
if [[ ! -f "$NODES_JSON" ]]; then
echo "[ERROR] nodes.json missing at $NODES_JSON" >&2
exit 1
fi
python3 - "$NODES_JSON" <<'PY'
import json, sys
with open(sys.argv[1]) as handle:
nodes = json.load(handle)
assert len(nodes) == 1, nodes
entry = nodes[0]
assert entry["node_id"], entry
PY
echo "[INFO] Master reflects agent health and nodes.json entries"

View File

@@ -0,0 +1,108 @@
#!/usr/bin/env bash
set -euo pipefail
SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
TEST_ROOT="$(cd "$SCRIPT_DIR/.." && pwd)"
TMP_ROOT="$TEST_ROOT/tmp"
API_BASE="http://localhost:32300/api/v1/master"
NODE_ID="$(cat "$TMP_ROOT/node_id")"
AGENT_HOSTNAME="dev-e2euser-e2einst-pod-0"
NETWORK_NAME="tests_default"
NEW_AGENT_IP="172.28.0.200"
compose() {
if docker compose version >/dev/null 2>&1; then
docker compose "$@"
else
docker-compose "$@"
fi
}
before_file="$TMP_ROOT/before_restart.json"
curl -sS "$API_BASE/nodes/$NODE_ID" -o "$before_file"
prev_last_updated=$(python3 - "$before_file" <<'PY'
import json, sys
with open(sys.argv[1]) as handle:
node = json.load(handle)
print(node.get("last_updated", ""))
PY
)
prev_ip=$(python3 - "$before_file" <<'PY'
import json, sys
with open(sys.argv[1]) as handle:
node = json.load(handle)
print(node["meta_data"].get("ip", ""))
PY
)
initial_ip=$(cat "$TMP_ROOT/initial_ip")
if [[ "$prev_ip" != "$initial_ip" ]]; then
echo "[ERROR] Expected initial IP $initial_ip, got $prev_ip" >&2
exit 1
fi
pushd "$TEST_ROOT" >/dev/null
compose rm -sf agent
popd >/dev/null
docker container rm -f argus-agent-e2e >/dev/null 2>&1 || true
AGENT_DIR="$TEST_ROOT/private/argus/agent/$AGENT_HOSTNAME"
HEALTH_DIR="$TEST_ROOT/private/argus/agent/health/$AGENT_HOSTNAME"
# 先以 sleep 方式启动容器,确保我们掌握注册时的网络状态
if ! docker run -d \
--name argus-agent-e2e \
--hostname "$AGENT_HOSTNAME" \
--network "$NETWORK_NAME" \
--ip "$NEW_AGENT_IP" \
-v "$AGENT_DIR:/private/argus/agent/$AGENT_HOSTNAME" \
-v "$HEALTH_DIR:/private/argus/agent/health/$AGENT_HOSTNAME" \
argus-agent:dev \
sleep 300 >/dev/null; then
echo "[ERROR] Failed to start agent container with custom IP" >&2
exit 1
fi
# 在容器内启动真实 agent 进程
if ! docker exec -d argus-agent-e2e python -m app.main; then
echo "[ERROR] Failed to spawn agent process inside container" >&2
exit 1
fi
success=false
detail_file="$TMP_ROOT/post_restart.json"
for _ in {1..20}; do
sleep 3
if ! curl -sS "$API_BASE/nodes/$NODE_ID" -o "$detail_file"; then
continue
fi
if python3 - "$detail_file" "$prev_last_updated" "$NODE_ID" "$prev_ip" "$NEW_AGENT_IP" <<'PY'
import json, sys
with open(sys.argv[1]) as handle:
node = json.load(handle)
prev_last_updated = sys.argv[2]
expected_id = sys.argv[3]
old_ip = sys.argv[4]
expected_ip = sys.argv[5]
last_updated = node.get("last_updated")
current_ip = node["meta_data"].get("ip")
assert node["id"] == expected_id
if current_ip != expected_ip:
raise SystemExit(1)
if current_ip == old_ip:
raise SystemExit(1)
if not last_updated or last_updated == prev_last_updated:
raise SystemExit(1)
PY
then
success=true
break
fi
done
if [[ "$success" != true ]]; then
echo "[ERROR] Agent did not report expected new IP $NEW_AGENT_IP after restart" >&2
exit 1
fi
echo "[INFO] Agent restart produced successful re-registration with IP change"

View File

@@ -0,0 +1,24 @@
#!/usr/bin/env bash
set -euo pipefail
SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
TEST_ROOT="$(cd "$SCRIPT_DIR/.." && pwd)"
compose() {
if docker compose version >/dev/null 2>&1; then
docker compose "$@"
else
docker-compose "$@"
fi
}
docker container rm -f argus-agent-e2e >/dev/null 2>&1 || true
pushd "$TEST_ROOT" >/dev/null
compose down --remove-orphans
popd >/dev/null
rm -rf "$TEST_ROOT/private"
rm -rf "$TEST_ROOT/tmp"
echo "[INFO] Agent E2E environment cleaned up"