fix(syslog-consumer-rule): 修复现场日志解析并补充升级工具
- 支持设备头字段值包含空格,避免 WAF 等名称被误拆为无效键值对 - 将空字符串 cropper offset 按未配置处理,兼容历史标准化规则 - 增加设备头解析和空 offset 回归测试 - 增加统计物化 Topic 创建脚本 - 增加容器内 JAR 热替换脚本及操作说明
This commit is contained in:
@@ -0,0 +1,75 @@
|
||||
#!/usr/bin/env bash
|
||||
set -Eeuo pipefail
|
||||
|
||||
# 现场临时热替换 syslog-consumer-rule 容器内的 JAR。
|
||||
# 默认容器名和容器内路径来自当前生产镜像:
|
||||
# syslog-consumer-rule-dm:/app/syslog-consumer-rule.jar
|
||||
#
|
||||
# 用法:
|
||||
# ./upgrade-doc/replace-syslog-consumer-rule-jar.sh
|
||||
# ./upgrade-doc/replace-syslog-consumer-rule-jar.sh /path/to/syslog-consumer-rule-1.0.0.jar
|
||||
|
||||
CONTAINER_NAME="syslog-consumer-rule-dm"
|
||||
CONTAINER_JAR="/app/syslog-consumer-rule.jar"
|
||||
|
||||
SCRIPT_DIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)
|
||||
PROJECT_DIR=$(cd "${SCRIPT_DIR}/.." && pwd)
|
||||
DEFAULT_JAR="${PROJECT_DIR}/syslog-consumer-rule/target/syslog-consumer-rule-1.0.0.jar"
|
||||
LOCAL_JAR="${1:-${DEFAULT_JAR}}"
|
||||
BACKUP_DIR="${SCRIPT_DIR}/backups"
|
||||
BACKUP_JAR="${BACKUP_DIR}/syslog-consumer-rule-$(date '+%Y%m%d-%H%M%S').jar"
|
||||
STARTED_AT=""
|
||||
|
||||
fail() {
|
||||
echo "ERROR: $*" >&2
|
||||
exit 1
|
||||
}
|
||||
|
||||
command -v docker >/dev/null 2>&1 || fail "未找到 docker 命令"
|
||||
[[ -f "${LOCAL_JAR}" && -s "${LOCAL_JAR}" ]] || fail "新 JAR 不存在或为空: ${LOCAL_JAR}"
|
||||
docker inspect "${CONTAINER_NAME}" >/dev/null 2>&1 || fail "容器不存在: ${CONTAINER_NAME}"
|
||||
|
||||
mkdir -p "${BACKUP_DIR}"
|
||||
|
||||
echo "[1/5] 备份容器内旧 JAR"
|
||||
docker cp "${CONTAINER_NAME}:${CONTAINER_JAR}" "${BACKUP_JAR}"
|
||||
echo "旧 JAR 已备份到: ${BACKUP_JAR}"
|
||||
|
||||
echo "[2/5] 停止容器 ${CONTAINER_NAME}"
|
||||
docker stop "${CONTAINER_NAME}" >/dev/null
|
||||
|
||||
restart_after_failure() {
|
||||
local exit_code=$?
|
||||
if [[ ${exit_code} -ne 0 ]]; then
|
||||
echo "替换过程失败,尝试重新启动容器。旧 JAR 备份位于: ${BACKUP_JAR}" >&2
|
||||
docker start "${CONTAINER_NAME}" >/dev/null 2>&1 || true
|
||||
fi
|
||||
exit "${exit_code}"
|
||||
}
|
||||
trap restart_after_failure ERR
|
||||
|
||||
echo "[3/5] 替换容器内 JAR"
|
||||
docker cp "${LOCAL_JAR}" "${CONTAINER_NAME}:${CONTAINER_JAR}"
|
||||
|
||||
echo "[4/5] 启动容器"
|
||||
STARTED_AT=$(date -u '+%Y-%m-%dT%H:%M:%SZ')
|
||||
docker start "${CONTAINER_NAME}" >/dev/null
|
||||
sleep 8
|
||||
|
||||
RUNNING=$(docker inspect --format '{{.State.Running}}' "${CONTAINER_NAME}")
|
||||
if [[ "${RUNNING}" != "true" ]]; then
|
||||
trap - ERR
|
||||
echo "ERROR: 容器启动后未保持运行,启动日志如下:" >&2
|
||||
docker logs --since "${STARTED_AT}" "${CONTAINER_NAME}" 2>&1 | tail -80 >&2 || true
|
||||
echo "可使用以下命令回滚:" >&2
|
||||
echo "docker cp '${BACKUP_JAR}' '${CONTAINER_NAME}:${CONTAINER_JAR}' && docker start '${CONTAINER_NAME}'" >&2
|
||||
exit 1
|
||||
fi
|
||||
|
||||
trap - ERR
|
||||
|
||||
echo "[5/5] 替换完成,显示本次启动日志"
|
||||
echo "新 JAR: ${LOCAL_JAR}"
|
||||
echo "旧 JAR 备份: ${BACKUP_JAR}"
|
||||
echo "注意:以后若删除或重新创建容器,容器内替换会丢失,仍需更新镜像。"
|
||||
docker logs --since "${STARTED_AT}" "${CONTAINER_NAME}" 2>&1 | tail -80
|
||||
@@ -0,0 +1,10 @@
|
||||
|
||||
|
||||
docker inspect syslog-consumer-rule-dm \
|
||||
--format 'WorkingDir={{.Config.WorkingDir}}{{println}}Entrypoint={{json .Config.Entrypoint}}{{println}}Cmd={{json .Config.Cmd}}'
|
||||
|
||||
|
||||
再确认文件确实存在:
|
||||
|
||||
docker exec syslog-consumer-rule-dm \
|
||||
sh -c 'ls -lh /app/*.jar 2>/dev/null'
|
||||
+120
@@ -0,0 +1,120 @@
|
||||
#!/usr/bin/env bash
|
||||
|
||||
set -euo pipefail
|
||||
|
||||
# 创建 syslog-consumer-rule 统计物化 Topic。
|
||||
# 默认值对应当前客户现场,直接执行即可:
|
||||
# bash 03-create-stats-materialize-topic.sh
|
||||
#
|
||||
# 其他环境可在执行前覆盖:BOOTSTRAP、RAW_TOPIC、STATS_TOPIC、KAFKA_BIN、COMMAND_CONFIG。
|
||||
|
||||
SCRIPT_DIR=$(cd -- "$(dirname -- "${BASH_SOURCE[0]}")" && pwd)
|
||||
BOOTSTRAP="${BOOTSTRAP:-10.141.121.35:9092,10.141.121.50:9092,10.141.121.59:9092}"
|
||||
RAW_TOPIC="${RAW_TOPIC:-agent-01-syslog-topic}"
|
||||
STATS_TOPIC="${STATS_TOPIC:-${RAW_TOPIC}.STATS.MATERIALIZE}"
|
||||
RETENTION_MS="${RETENTION_MS:-604800000}"
|
||||
|
||||
if [[ -z "${KAFKA_BIN:-}" ]]; then
|
||||
for CANDIDATE_BIN in \
|
||||
"${SCRIPT_DIR}" \
|
||||
"/opt/kafka/kafka_2.13-3.6.1/bin" \
|
||||
"/opt/kafka/bin"; do
|
||||
if [[ -x "${CANDIDATE_BIN}/kafka-topics.sh" ]] \
|
||||
&& [[ -x "${CANDIDATE_BIN}/kafka-configs.sh" ]]; then
|
||||
KAFKA_BIN="${CANDIDATE_BIN}"
|
||||
break
|
||||
fi
|
||||
done
|
||||
fi
|
||||
|
||||
if [[ -z "${KAFKA_BIN:-}" ]] && command -v kafka-topics.sh >/dev/null 2>&1; then
|
||||
KAFKA_BIN=$(dirname -- "$(command -v kafka-topics.sh)")
|
||||
fi
|
||||
|
||||
if [[ -z "${KAFKA_BIN:-}" ]]; then
|
||||
echo "ERROR: 未找到 Kafka CLI;现场预期路径为 /opt/kafka/kafka_2.13-3.6.1/bin" >&2
|
||||
exit 1
|
||||
fi
|
||||
|
||||
TOPICS_CMD="${KAFKA_BIN}/kafka-topics.sh"
|
||||
CONFIGS_CMD="${KAFKA_BIN}/kafka-configs.sh"
|
||||
COMMON_ARGS=(--bootstrap-server "${BOOTSTRAP}")
|
||||
|
||||
if [[ -n "${COMMAND_CONFIG:-}" ]]; then
|
||||
if [[ ! -r "${COMMAND_CONFIG}" ]]; then
|
||||
echo "ERROR: 无法读取认证配置 ${COMMAND_CONFIG}" >&2
|
||||
exit 1
|
||||
fi
|
||||
COMMON_ARGS+=(--command-config "${COMMAND_CONFIG}")
|
||||
fi
|
||||
|
||||
echo "Kafka:${BOOTSTRAP}"
|
||||
echo "原始 Topic:${RAW_TOPIC}"
|
||||
echo "统计 Topic:${STATS_TOPIC}"
|
||||
|
||||
echo "[1/4] 读取原始 Topic 的分区数和副本数"
|
||||
RAW_INFO=$("${TOPICS_CMD}" "${COMMON_ARGS[@]}" --describe --topic "${RAW_TOPIC}")
|
||||
printf '%s\n' "${RAW_INFO}"
|
||||
|
||||
PARTITIONS=$(printf '%s\n' "${RAW_INFO}" | awk '
|
||||
/PartitionCount:/ {
|
||||
for (i=1; i<=NF; i++) if ($i=="PartitionCount:") { print $(i+1); exit }
|
||||
}
|
||||
')
|
||||
REPLICAS=$(printf '%s\n' "${RAW_INFO}" | awk '
|
||||
/ReplicationFactor:/ {
|
||||
for (i=1; i<=NF; i++) if ($i=="ReplicationFactor:") { print $(i+1); exit }
|
||||
}
|
||||
')
|
||||
|
||||
if [[ ! "${PARTITIONS}" =~ ^[1-9][0-9]*$ ]] || [[ ! "${REPLICAS}" =~ ^[1-9][0-9]*$ ]]; then
|
||||
echo "ERROR: 无法读取 ${RAW_TOPIC} 的分区数或副本数" >&2
|
||||
exit 1
|
||||
fi
|
||||
echo "将沿用:分区数=${PARTITIONS},副本数=${REPLICAS}"
|
||||
|
||||
echo "[2/4] 创建统计 Topic(已存在时不重复创建)"
|
||||
"${TOPICS_CMD}" "${COMMON_ARGS[@]}" \
|
||||
--create --if-not-exists \
|
||||
--topic "${STATS_TOPIC}" \
|
||||
--partitions "${PARTITIONS}" \
|
||||
--replication-factor "${REPLICAS}" \
|
||||
--config cleanup.policy=delete \
|
||||
--config "retention.ms=${RETENTION_MS}"
|
||||
|
||||
echo "[3/4] 统一保留策略"
|
||||
if ! "${CONFIGS_CMD}" "${COMMON_ARGS[@]}" \
|
||||
--alter --entity-type topics --entity-name "${STATS_TOPIC}" \
|
||||
--add-config "cleanup.policy=delete,retention.ms=${RETENTION_MS}"; then
|
||||
echo "WARN: Kafka 配置更新请求超时或连接中断,将继续验证 Topic。" >&2
|
||||
echo "WARN: 新建命令已经携带相同配置,不影响 syslog-consumer-rule 使用该 Topic。" >&2
|
||||
fi
|
||||
|
||||
echo "[4/4] 验证统计 Topic"
|
||||
STATS_INFO=$("${TOPICS_CMD}" "${COMMON_ARGS[@]}" --describe --topic "${STATS_TOPIC}")
|
||||
printf '%s\n' "${STATS_INFO}"
|
||||
|
||||
ACTUAL_PARTITIONS=$(printf '%s\n' "${STATS_INFO}" | awk '
|
||||
/PartitionCount:/ {
|
||||
for (i=1; i<=NF; i++) if ($i=="PartitionCount:") { print $(i+1); exit }
|
||||
}
|
||||
')
|
||||
ACTUAL_REPLICAS=$(printf '%s\n' "${STATS_INFO}" | awk '
|
||||
/ReplicationFactor:/ {
|
||||
for (i=1; i<=NF; i++) if ($i=="ReplicationFactor:") { print $(i+1); exit }
|
||||
}
|
||||
')
|
||||
|
||||
if [[ "${ACTUAL_PARTITIONS}" != "${PARTITIONS}" ]]; then
|
||||
echo "ERROR: ${STATS_TOPIC} 已存在,但分区数=${ACTUAL_PARTITIONS},与原始 Topic 的 ${PARTITIONS} 不一致" >&2
|
||||
echo "为避免同一 key 重新映射,脚本不会自动增加分区,请先联系研发确认。" >&2
|
||||
exit 1
|
||||
fi
|
||||
if [[ "${ACTUAL_REPLICAS}" != "${REPLICAS}" ]]; then
|
||||
echo "ERROR: ${STATS_TOPIC} 副本数=${ACTUAL_REPLICAS},预期=${REPLICAS}" >&2
|
||||
exit 1
|
||||
fi
|
||||
|
||||
echo "PASS: ${STATS_TOPIC} 已创建并验证成功。"
|
||||
echo "分区数=${ACTUAL_PARTITIONS},副本数=${ACTUAL_REPLICAS},保留时间=${RETENTION_MS}ms。"
|
||||
echo "脚本未删除消息,未修改任何消费组 offset。"
|
||||
@@ -163,7 +163,7 @@ services:
|
||||
## 发布顺序
|
||||
|
||||
升级 `syslog-consumer-rule` 前,先在 Kafka 集群执行一次
|
||||
`upgrade-doc/v1/kafka/kafka-cluster-upgrade.sh`,确认已经创建:
|
||||
`upgrade-doc/v1/kafka/03-create-stats-materialize-topic.sh`,确认已经创建:
|
||||
|
||||
```text
|
||||
agent-01-syslog-topic.STATS.MATERIALIZE
|
||||
|
||||
Reference in New Issue
Block a user