1
k8s单个pod资源限制
在比格云主机上通过docker-compose管理的容器
华熙生物-300人用
1
2
3
4
5
6
7
8
9
| NAME CPU % MEM USAGE / LIMIT MEM %
hxsw-todo-1 0.17% 124.3MiB / 3.81GiB 3.19%
hxsw-lowcode-1 0.71% 972.5MiB / 3.81GiB 24.93%
hxsw-oaa-1 0.07% 190.9MiB / 3.81GiB 4.89%
hxsw-timejob-1 0.39% 81.33MiB / 3.81GiB 2.08%
hxsw-nginx-1 0.00% 52.87MiB / 3.81GiB 1.36%
hxsw-pointer-1 0.08% 38.64MiB / 3.81GiB 0.99%
hxsw-inroad-1 0.07% 503.3MiB / 3.81GiB 12.90%
合计 1.49% 1961MB
|
健坤化学，250人用量
1
2
3
4
5
6
7
8
9
| NAME CPU % MEM USAGE / LIMIT MEM % NET I/O BLOCK I/O PIDS
jkhx-inroad-1 1.49% 767.1MiB / 3.81GiB 19.66% 6.8GB / 13.1GB 25.7MB / 271MB 45
jkhx-nginx-1 0.00% 94.47MiB / 3.81GiB 2.42% 2.95GB / 16.3GB 247MB / 8.19kB 3
jkhx-lowcode-1 0.48% 356.1MiB / 3.81GiB 9.13% 6.69GB / 8.6GB 70.2MB / 0B 44
jkhx-timejob-1 0.12% 89.02MiB / 3.81GiB 2.28% 1.7GB / 1.25GB 9.97MB / 0B 23
jkhx-oaa-1 0.09% 216.9MiB / 3.81GiB 5.56% 227MB / 252MB 37MB / 19.3MB 21
jkhx-todo-1 0.20% 170MiB / 3.81GiB 4.36% 444MB / 606MB 51.9MB / 0B 23
jkhx-pointer-1 0.08% 30.17MiB / 3.81GiB 0.77% 438kB / 0B 17.4MB / 0B 14
合计 2.5% 1723
|
福邦新材料
1
2
3
4
5
6
7
8
| NAME CPU % MEM USAGE / LIMIT MEM % NET I/O BLOCK I/O PIDS
pointer 0.13% 35.14MiB / 3.682GiB 0.93% 76.4kB / 0B 55.6MB / 0B 15
timejob 0.08% 45.75MiB / 3.682GiB 1.21% 84.4MB / 91.6MB 8.76MB / 0B 19
oaa 0.08% 200MiB / 3.682GiB 5.31% 364MB / 354MB 4.97MB / 0B 21
inroad 0.26% 1.212GiB / 3.682GiB 32.93% 39.9GB / 39.1GB 33.6MB / 0B 44
nginx 0.00% 46.57MiB / 3.682GiB 1.24% 11.6GB / 113GB 50.1MB / 30.7kB 2
todo 0.19% 285.6MiB / 3.682GiB 7.58% 309MB / 309MB 32.5MB / 0B 21
总计 0.74% 1854
|
gc上通过k8s管理的容器（单pod多容器）
bach正式环境，300人规模
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
| POD NAME CPU(cores) MEMORY(bytes)
inroadweb-bach-768565bfc7-n24jh inroad 8m 561Mi
inroadweb-bach-768565bfc7-n24jh lowcode 111m 1073Mi
inroadweb-bach-768565bfc7-n24jh nginx 1m 6Mi
inroadweb-bach-768565bfc7-n24jh oaa 1m 304Mi
inroadweb-bach-768565bfc7-n24jh pointer 1m 34Mi
inroadweb-bach-768565bfc7-n24jh promtail 6m 48Mi
总计 126m 2028Mi
POD NAME CPU(cores) MEMORY(bytes)
inroadweb-bach-768565bfc7-q8g57 inroad 3m 517Mi
inroadweb-bach-768565bfc7-q8g57 lowcode 20m 1721Mi
inroadweb-bach-768565bfc7-q8g57 nginx 1m 16Mi
inroadweb-bach-768565bfc7-q8g57 oaa 9m 238Mi
inroadweb-bach-768565bfc7-q8g57 pointer 1m 50Mi
inroadweb-bach-768565bfc7-q8g57 promtail 4m 43Mi
总计 35m 2588Mi
|
zjcm正式环境
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
|
POD NAME CPU(cores) MEMORY(bytes)
inroadweb-zjcm-78d7ddc9bf-mpmjq inroad 2m 432Mi
inroadweb-zjcm-78d7ddc9bf-mpmjq lowcode 2m 1100Mi
inroadweb-zjcm-78d7ddc9bf-mpmjq nginx 1m 8Mi
inroadweb-zjcm-78d7ddc9bf-mpmjq oaa 1m 245Mi
inroadweb-zjcm-78d7ddc9bf-mpmjq pointer 1m 69Mi
inroadweb-zjcm-78d7ddc9bf-mpmjq promtail 2m 47Mi
合计 9m 1900Mi
inroadweb-zjcm-78d7ddc9bf-8mcxk inroad 2m 496Mi
inroadweb-zjcm-78d7ddc9bf-8mcxk lowcode 3m 1441Mi
inroadweb-zjcm-78d7ddc9bf-8mcxk nginx 1m 4Mi
inroadweb-zjcm-78d7ddc9bf-8mcxk oaa 1m 259Mi
inroadweb-zjcm-78d7ddc9bf-8mcxk pointer 1m 62Mi
inroadweb-zjcm-78d7ddc9bf-8mcxk promtail 5m 36Mi
合计 9m 2300Mi
|
bach测试环境，没几个人用
1
2
3
4
5
6
7
| POD NAME CPU(cores) MEMORY(bytes)
inroadweb-bach-68d88fdbcb-2dh2c inroad 2m 381Mi
inroadweb-bach-68d88fdbcb-2dh2c lowcode 4m 721Mi
inroadweb-bach-68d88fdbcb-2dh2c nginx 0m 9Mi
inroadweb-bach-68d88fdbcb-2dh2c oaa 1m 199Mi
inroadweb-bach-68d88fdbcb-2dh2c pointer 1m 51Mi
总计 7m 1363Mi
|
zjcm测试环境，没几个人用
1
2
3
4
5
6
7
8
| POD NAME CPU(cores) MEMORY(bytes)
inroadweb-zjcm-7c97cfbdf8-lfjx7 inroad 5m 336Mi
inroadweb-zjcm-7c97cfbdf8-lfjx7 lowcode 3m 230Mi
inroadweb-zjcm-7c97cfbdf8-lfjx7 nginx 1m 6Mi
inroadweb-zjcm-7c97cfbdf8-lfjx7 oaa 1m 172Mi
inroadweb-zjcm-7c97cfbdf8-lfjx7 pointer 1m 50Mi
inroadweb-zjcm-7c97cfbdf8-lfjx7 promtail 10m 56Mi
合计 19m 853Mi
|
比格云测试集群（多pod多容器）
xsd，0人使用
1
2
3
4
5
6
7
8
9
| NAME CPU % MEM USAGE
hxsw-todo-1 0 115
hxsw-lowcode-1 0 254
hxsw-oaa-1 0 173
hxsw-timejob-1 0 106
hxsw-nginx-1 0 4
hxsw-pointer-1 0 33
hxsw-inroad-1 0 393
合计 0 1078
|
结论
:star:无论时单容器但是多容器，创建一套服务的成本大概是1G内存，用于deployment的资源请求配额
:star:内存占用排序：积木、工智道、oaa、todo、pointer、timejob、nginx
| 服务 |
inroad |
lowcode |
oaa |
todo |
pointer |
timejob |
nginx |
deployment |
| 成本 |
400MB |
300M |
180M |
120M |
50M |
120M |
10M |
1200M |
| 峰值 |
1241M |
1800M |
360M |
300M |
100M |
none |
none |
|
CPU 请求：200m = 0.2 核心
资源request
1
2
3
4
| resources:
requests:
cpu: 200m
memory: 256Mi
|
弹性伸缩规则
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
| apiVersion: autoscaling/v2beta2
kind: HorizontalPodAutoscaler
metadata:
name: my-hpa
spec:
scaleTargetRef:
apiVersion: apps/v1
kind: Deployment
name: my-deployment
minReplicas: 1
maxReplicas: 5
metrics:
- type: Resource
resource:
name: memory
target:
type: Utilization
averageUtilization: 70
|
inroad
工智道随便开几个页面访问，内存就能达到一个G
要求，大概没人访问的时候，大概占用内存400M，只有一个pod，请求2个G，达到80%扩容新的pod
1
2
3
4
| resources:
requests:
cpu: 200m
memory: 1200Mi
|
收集dump文件
1
2
| https://zhuanlan.zhihu.com/p/639610440
https://learn.microsoft.com/zh-cn/dotnet/core/diagnostics/dotnet-dump
|
1
2
3
4
5
6
7
| #!/bin/bash
mkdir /opt/$1 $$ cd /opt/$1
unzip config.zip
wget -q https://3dview-1251252938.cos.ap-shanghai.myqcloud.com/dengjinjun/scripts/install_docker.sh
chmod +x install_docker.sh
sh install_docker.sh
docker-compose up -d
|
