背景

在日常使用 efk 日志系统的过程中,每天造成的日志索引量是巨大的,需要进行对相关的索引进行定期清理,来缓解后端储存的占用。如何实现在 kubernetes 中进行优雅的定期清理?本文将介绍使用 kubernetes 中的 Cronjob 资源对象进行其需求的实现。

常规情况下清理 n day 以前的数据

1
curl -XDELETE http://elasticsearch-logging.kube-system:9200/logstash-`date -d"n days ago" +"%Y.%m.%d"`

这种方法,是网上比较常规的方式,如某一天忘记执行了,就会出现数据未清理干净的情况,有点不优雅,如何解决?

进行优化后的 清理策略

1
2
3
4
5
6
7
8
9
cleanPrefixPath='k8s-'
agoCleanTime=`date -d "n days ago" +%s`

for i in `curl http://user:pass@es_url/_cat/indices?v 2>&1 |awk -F '[ ]+' '{print $3}'|sort -n |grep -v '^\.'|egrep "${cleanPrefixPath}[0-9]+.[0-9]+.[0-9]+"`;do
        secTime=`echo ${i#*-}|tr '.' '-'|xargs -I {} date -d "{}" +%s`
        if [ "${secTime}" -le "${agoCleanTime}" ];then
                curl -XDELETE http://user:pass@es_url/"${i}"
        fi
done

转换为Yaml资源清单后呢?

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
# es-clean-cronjob.yaml
apiVersion: v1
kind: ConfigMap
metadata:
  name: es-auth-up
data:
  user: "elastic"
  password: "xxxxx"
  es_url: "elasticsearch-data.logging:9200"
  
---
apiVersion: batch/v1beta1 # or batch/v1
kind: CronJob
metadata:
  name: es-clean-job
spec:
  schedule: "0 12 * * *" # 每天中午 12:00 执行
  jobTemplate:
    spec:
      template:
        spec:
          containers:
          - name: es-clean-job
            image: idocker.io/centos:7
            imagePullPolicy: IfNotPresent
            env: 
              - name: ES_USER
                valueFrom:
                  configMapKeyRef:
                    name: es-auth-up
                    key: user
              - name: ES_PASS
                valueFrom:
                  configMapKeyRef:
                    name: es-auth-up
                    key: password
              - name: ES_URL
                valueFrom:
                  configMapKeyRef:
                    name: es-auth-up
                    key: es_url
            command: ["/bin/bash"]
            args:
              - -c
              - |
                set -eu
                ln -sf /usr/share/zoneinfo/Asia/Shanghai /etc/localtime
                cleanPrefixPath='k8s-'  # 删除前缀, 设置为空时将对所有索引生效
                agoCleanTime=`date -d "30 days ago" +%s` # 删除 30 天以前的日志
                for i in `curl http://"${ES_USER}":"${ES_PASS}"@"${ES_URL}"/_cat/indices?v 2>&1 |awk -F '[ ]+' '{print $3}'|sort -n |grep -v '^\.'|egrep "${cleanPrefixPath}[0-9]+.[0-9]+.[0-9]+"`;do
                        secTime=`echo ${i#*-}|tr '.' '-'|xargs -I {} date -d "{}" +%s`
                        if [ "${secTime}" -le "${agoCleanTime}" ];then
                                curl -XDELETE http://"${ES_USER}":"${ES_PASS}"@"${ES_URL}"/"${i}"
                        fi
                done                
          restartPolicy: Never
      backoffLimit: 2

执行创建

1
kubectl apply -f ./es-clean-cronjob.yaml  -n logging 

示例的索引结构

image-20210713121202811

执行时的日志输出

image-20210713120526442

问题记录

ToDo