kdoctor agent schedule #137

Icarus9913 · 2023-08-11T06:30:12Z

No description provided.

codecov · 2023-08-11T06:33:29Z

Codecov Report

Merging #137 (c5368a7) into main (77b1679) will decrease coverage by 1.69%.
Report is 6 commits behind head on main.
The diff coverage is 0.00%.

@@            Coverage Diff             @@
##             main     #137      +/-   ##
==========================================
- Coverage   40.63%   38.94%   -1.69%     
==========================================
  Files           8        8              
  Lines         507      529      +22     
==========================================
  Hits          206      206              
- Misses        296      318      +22     
  Partials        5        5

Flag	Coverage Δ
unittests	`38.94% <0.00%> (-1.69%)`	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

Files Changed	Coverage Δ
pkg/reportManager/manager.go	`0.00% <0.00%> (ø)`
pkg/reportManager/worker.go	`17.69% <0.00%> (-3.41%)`	⬇️

Signed-off-by: Icarus9913 <icaruswu66@qq.com>

weizhoublue · 2023-08-15T03:24:58Z

test/docs/Runtime.md

+
+| Case ID | Title                                                             | Priority | Smoke | Status |    Other    |
+|---------|-------------------------------------------------------------------|----------|-------|--------|-------------|
+| E00001  | Successfully testing Task Runtime creation                        | p1       |       |        |             |


这个描述可以更细化下，本身就是希望看描述知道有了哪些用例

例如，它是建立哪个 CRD 还是所有？它是建立 deployment 还是 daemonset ？

这个用例有补充，还在写没有提上来

weizhoublue · 2023-08-15T03:27:33Z

test/docs/Runtime.md

+|---------|-------------------------------------------------------------------|----------|-------|--------|-------------|
+| E00001  | Successfully testing Task Runtime creation                        | p1       |       |        |             |
+| E00002  | Successfully testing Task Runtime Service creation                | p1       |       |        |             |
+| E00003  | Successfully testing cascading deletion with Task Runtime Service | p1       |       |        |             |


BTW，有没有用例校验，包括资源的创建、删除时间、status 中间状态的装换等符合 spec 中的预期？

这个用例有补充，还在写没有提上来

weizhoublue · 2023-08-15T03:30:17Z

docs/reference/apphttphealthy-zh_CN.md

+
+| 字段                            | 描述                                       | 结构                                                                                                                                     | 验证  | 取值                   | 默认值       |
+|-------------------------------|------------------------------------------|----------------------------------------------------------------------------------------------------------------------------------------|-----|----------------------|-----------|
+| annotation                    | agent 工作负载的 annotation 配合搭配 multus 多网卡使用 | map[string]string                                                                                                                      | 可选  |                      |           |


配合搭配 multus 多网卡使用
这个就不要限定误导

weizhoublue · 2023-08-15T03:33:01Z

docs/usage/apphttphealthy-zh_CN.md

@@ -4,7 +4,7 @@

 ## 介绍 

-对于这种任务，每个 kdoctor agent 都会向指定的目标发送 http 请求，默认并发量为 50 可覆盖多副本情况，并发量可在 kodcotr 的 configmap 中设置，并获得成功率和平均延迟。根据成功条件来判断结果是否成功。并且，可以通过聚合API获取详细的报告。
+对于这种任务， kdoctor-controller 会根据 agentSpec 生成对应的 agent ，每一个 agent pod 都会向指定的目标发送 http 请求，默认并发量为 50 可覆盖多副本情况，并发量可在 kodcotr 的 configmap 中设置，并获得成功率和平均延迟。根据成功条件来判断结果是否成功。并且，可以通过聚合API获取详细的报告。



资源的创建，包括哪些？ deloyment 和 service
资源的删除逻辑是什么？
报告的收取的逻辑是什么？删除后是否影响报告？deployment 删除和 CR 删除，和报告保留是什么关系

建议单独用一个 md 说明这些，每个 CRD 中使用引用跳转

weizhoublue · 2023-08-15T03:34:30Z

docs/usage/apphttphealthy-zh_CN.md

@@ -67,6 +67,10 @@ kind: AppHttpHealthy
 metadata:
  name: http1
 spec:
+  agentSpec:


get-started就是要精简，按默认工作即可，不要写这些

weizhoublue · 2023-08-15T03:36:17Z

docs/reference/netdns-zh_CN.md

+| env                           | agent 工作负载环境变量                           | [env](https://github.com/kubernetes/kubernetes/blob/v1.27.0/staging/src/k8s.io/api/core/v1/types.go#L2012)                             | 可选  |                      |           |
+| hostNetwork                   | agent 工作负载是否使用宿主机网络                      | bool                                                                                                                                   | 可选  | true、false           | false     |
+| resources                     | agent 工作负载资源使用配置                         | [resources](https://github.com/kubernetes/kubernetes/blob/v1.27.0/staging/src/k8s.io/api/core/v1/types.go#L2333)             | 可选  |                      |         |
+| terminationGracePeriodMinutes | agent 工作负载完成任务后多少分钟之后终止                  | int                                                                                                                                    | 可选  | 大于等于 0               | 60        |


这个默认值 60 在 chart values 中可定制，更合适

weizhoublue · 2023-08-15T03:36:39Z

docs/reference/netdns-zh_CN.md

+| affinity                      | agent 工作负载亲和性                            | [labelSelector](https://github.com/kubernetes/kubernetes/blob/v1.27.0/staging/src/k8s.io/apimachinery/pkg/apis/meta/v1/types.go#L1195) | 可选  |                      |           |
+| env                           | agent 工作负载环境变量                           | [env](https://github.com/kubernetes/kubernetes/blob/v1.27.0/staging/src/k8s.io/api/core/v1/types.go#L2012)                             | 可选  |                      |           |
+| hostNetwork                   | agent 工作负载是否使用宿主机网络                      | bool                                                                                                                                   | 可选  | true、false           | false     |
+| resources                     | agent 工作负载资源使用配置                         | [resources](https://github.com/kubernetes/kubernetes/blob/v1.27.0/staging/src/k8s.io/api/core/v1/types.go#L2333)             | 可选  |                      |         |


默认是什么？无资源限制

默认cpu:100m.mem: 128Mi
在 chart value 中可以设置

我指文档要补充，不是 comment 中告知我

weizhoublue · 2023-08-15T03:55:24Z

test/docs/Runtime.md

+| E00001  | Successfully testing Task Runtime creation                        | p1       |       |        |             |
+| E00002  | Successfully testing Task Runtime Service creation                | p1       |       |        |             |
+| E00003  | Successfully testing cascading deletion with Task Runtime Service | p1       |       |        |             |
+| E00004  | Successfully testing cascading deletion with Task Runtime         | p1       |       |        |             |


（1）这些资源如果中途被人删除了会发生什么？业务代码是否会崩，CRD status 会展示什么，是否要加入 finalizer 或者 webhook 防范
（2）中途删除 CRD，期待什么，是否有用例

有用例的

weizhoublue · 2023-08-15T03:57:21Z

test/docs/Runtime.md

+| Case ID | Title                                                             | Priority | Smoke | Status |    Other    |
+|---------|-------------------------------------------------------------------|----------|-------|--------|-------------|
+| E00001  | Successfully testing Task Runtime creation                        | p1       |       |        |             |
+| E00002  | Successfully testing Task Runtime Service creation                | p1       |       |        |             |


项目卸载时，卸载流程是否有影响，是否有残余CR或 deployment

此功能暂时还没实现，等后面在做这个功能吧

这个很简单，spidernet-io/spiderpool@b5b8919
参考下，一切做了，或则要在文档中体现用例

这个 pr 里的东西太多了重新搞个 pr 搞吧

Signed-off-by: ii2day <ji.li@daocloud.io>

weizhoublue · 2023-08-22T03:27:33Z

charts/templates/deployment.yaml

@@ -73,6 +73,11 @@ spec:
            - {{ .Values.kdoctorController.cmdBinName }}
          args:
            - --config-path=/tmp/config-map/conf.yml
+            - --configmap-deployment-template=/tmp/configmap-app-template/deployment.yml


这个成本是否有点高，以后加一个配置就要写个命令行参数

这里是否给个 configmap name ，代码 get 自己去读
或者给个路径即可

代码量一样，全部扔到cofigmap里，后台代码读取后判断是否有值，然后再json.unmarshal给某个结构体实例，以后加一个配置，一样修改后台代码。

另此处，后台代码已做模版化处理，“验证，读取”全部流程化。

weizhoublue · 2023-08-22T03:30:52Z

docs/reference/apphttphealthy-zh_CN.md

+| kind                          | agent 工作负载的类型          | string                                                                                                                                 | 可选  | Deployment、DaemonSet | DaemonSet          |
+| deploymentReplicas            | agent 工作负载类型为 deployment 时的期望副本数 | int                                                                                                                                    | 可选  | 大于等于 0               | 0                  |
+| affinity                      | agent 工作负载亲和性          | [labelSelector](https://github.com/kubernetes/kubernetes/blob/v1.27.0/staging/src/k8s.io/apimachinery/pkg/apis/meta/v1/types.go#L1195) | 可选  |                      |                    |
+| env                           | agent 工作负载环境变量         | [env](https://github.com/kubernetes/kubernetes/blob/v1.27.0/staging/src/k8s.io/api/core/v1/types.go#L2012)                             | 可选  |                      |                    |


不能贴代码，代码是会变动了，行会变化的
如果需要用户感知，那就需要有个 referent/agent.md 说明启动命令和环境变量等

我看 spiderpool 中就是这么写的

https://kubernetes.io/docs/reference/generated/kubernetes-api/v1.23/#labelselector-v1-meta

这一样有版本问题

我们是小项目，随着代码迭代，行数变化，我们没有精力长期变更这个链接，也没 CI check 这个链接行数是否正确。
并且代码也没对环境变量的文字说明，没人看得懂这些是什么意思

参考 https://github.com/spidernet-io/spiderpool/blob/main/docs/reference/spiderpool-agent.md

Signed-off-by: ii2day <ji.li@daocloud.io>

weizhoublue · 2023-08-22T09:51:57Z

之前谈到，务必最好给 agent 设置 limit 资源，避免影响生产环境
chart values 可定义，有默认值

weizhoublue · 2023-08-22T09:53:21Z

docs/reference/runtime-zh_CN.md

+
+### 工作负载
+
+工作负载为 DaemonSet 或 Deployment，默认为 Daemonset，负载中的每一个 Pod 根据任务配置进行的请求，并将执行结果落盘到 Pod 中，可通过 AgentSpec 中设置


只有所有 pod 就绪了，任务才开始按照 spec 中的时间定义启动

weizhoublue · 2023-08-22T09:55:07Z

docs/reference/runtime-zh_CN.md

+的销毁逻辑相同。
+
+### Ingress
+


（1）关于任务CR 的删除、资源的优雅删除、报告的删除，三者间是什么关系，是否有个时序图之类的表达关系，运维才知道它的操作有什么影响
什么时候删除资源，什么时候删除 CR 是安全的

（2）任务的资源优雅删除是怎么设计的，为什么需要这个

ii2day · 2023-08-23T03:36:44Z

之前谈到，务必最好给 agent 设置 limit 资源，避免影响生产环境 chart values 可定义，有默认值

done

weizhoublue · 2023-08-23T07:18:04Z

docs/reference/runtime-zh_CN.md

@@ -0,0 +1,59 @@
+## runtime
+
+当下发任务 CR 后，kdoctor-controller 会根据 CR 中的 AgentSpec 生成对应的任务载体（DaemonSet 或 Deployment）当所有 Pod 就绪后，开始按照 Spec 中的任务定义执行任务，每一个任务独立使用一个载体。


这个文件在 doc/mkdoc 没有链接

weizhoublue · 2023-08-23T07:21:12Z

docs/reference/runtime-zh_CN.md

+    workload ->>ingress:  到达 runtime 销毁时间，销毁 ingress
+    cr 任务 ->>kdoctor_controller: cr 任务删除
+    kdoctor_controller ->> workload: cr 任务删除，workload 删除
+    workload ->> pod: workload 删除，pod 删除


（1）这个时序图中，好像还没说清报告的生命周期
（2）这个图的后续，是否可以给几个简单的结论，
报告的什么周期是什么（删除CR 是否意味着它的报告也会被删除了）

weizhoublue · 2023-08-23T10:41:39Z

docs/mkdocs.yml

@@ -50,6 +50,7 @@ nav:
      - AppHttpHealthy: reference/apphttphealthy.md
      - NetReach: reference/netreach.md
      - NetDns: reference/netdns.md
+      - Runtime: reference/runtime.md


这是 concept 章节更合适吧？

done 已更改

weizhoublue · 2023-08-24T04:07:12Z

修复下 ci

Signed-off-by: ii2day <ji.li@daocloud.io>

Icarus9913 added the release/feature-new label Aug 11, 2023

Icarus9913 requested a review from weizhoublue as a code owner August 11, 2023 06:30

Icarus9913 force-pushed the feat/wk/schedule-v2 branch 2 times, most recently from 1240af1 to 2a64ca0 Compare August 11, 2023 06:38

Icarus9913 changed the title ~~Feat/wk/schedule v2~~ kdoctor agent schedule Aug 11, 2023

agent scheduler

3f7f778

Signed-off-by: Icarus9913 <icaruswu66@qq.com>

Icarus9913 force-pushed the feat/wk/schedule-v2 branch from 2a64ca0 to 0ef7d92 Compare August 11, 2023 08:06

ii2day force-pushed the feat/wk/schedule-v2 branch 2 times, most recently from 40f446f to 4e8ebb7 Compare August 11, 2023 09:55

Icarus9913 force-pushed the feat/wk/schedule-v2 branch from 4e8ebb7 to 87becb5 Compare August 14, 2023 02:33

ii2day force-pushed the feat/wk/schedule-v2 branch from 87becb5 to 498f1f1 Compare August 14, 2023 06:51

weizhoublue reviewed Aug 15, 2023

View reviewed changes

ii2day force-pushed the feat/wk/schedule-v2 branch 2 times, most recently from 7f0f50e to 925f202 Compare August 15, 2023 06:05

Adapt scheduler report

afe5513

Signed-off-by: ii2day <ji.li@daocloud.io>

ii2day force-pushed the feat/wk/schedule-v2 branch 6 times, most recently from 7935ea4 to 9002712 Compare August 16, 2023 03:45

ii2day force-pushed the feat/wk/schedule-v2 branch 2 times, most recently from cb927e6 to 0738421 Compare August 16, 2023 10:32

weizhoublue reviewed Aug 22, 2023

View reviewed changes

ii2day force-pushed the feat/wk/schedule-v2 branch from 0738421 to ed7cb4e Compare August 22, 2023 07:51

add schedule e2e

52b93dc

Signed-off-by: ii2day <ji.li@daocloud.io>

ii2day force-pushed the feat/wk/schedule-v2 branch 3 times, most recently from 7a2ba07 to daab7b3 Compare August 22, 2023 09:08

weizhoublue reviewed Aug 22, 2023

View reviewed changes

ii2day force-pushed the feat/wk/schedule-v2 branch from daab7b3 to 276d452 Compare August 23, 2023 03:34

ii2day force-pushed the feat/wk/schedule-v2 branch 4 times, most recently from 4967054 to 5a453d0 Compare August 23, 2023 06:43

weizhoublue reviewed Aug 23, 2023

View reviewed changes

ii2day force-pushed the feat/wk/schedule-v2 branch from 5a453d0 to 18de65d Compare August 23, 2023 07:42

weizhoublue reviewed Aug 23, 2023

View reviewed changes

ii2day force-pushed the feat/wk/schedule-v2 branch from 18de65d to 2753559 Compare August 23, 2023 10:51

weizhoublue closed this Aug 23, 2023

weizhoublue reopened this Aug 23, 2023

ii2day force-pushed the feat/wk/schedule-v2 branch from 2753559 to b8e45e8 Compare August 24, 2023 06:47

update crd

c5368a7

Signed-off-by: ii2day <ji.li@daocloud.io>

ii2day force-pushed the feat/wk/schedule-v2 branch from b8e45e8 to c5368a7 Compare August 24, 2023 07:01

weizhoublue merged commit 27d2481 into main Aug 24, 2023
24 of 26 checks passed


		### 工作负载

		工作负载为 DaemonSet 或 Deployment，默认为 Daemonset，负载中的每一个 Pod 根据任务配置进行的请求，并将执行结果落盘到 Pod 中，可通过 AgentSpec 中设置

		@@ -0,0 +1,59 @@
		## runtime

		当下发任务 CR 后，kdoctor-controller 会根据 CR 中的 AgentSpec 生成对应的任务载体（DaemonSet 或 Deployment）当所有 Pod 就绪后，开始按照 Spec 中的任务定义执行任务，每一个任务独立使用一个载体。

kdoctor agent schedule #137

kdoctor agent schedule #137

Conversation

Icarus9913 commented Aug 11, 2023

codecov bot commented Aug 11, 2023 • edited Loading

Codecov Report

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

weizhoublue Aug 15, 2023 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

weizhoublue Aug 15, 2023 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

weizhoublue Aug 22, 2023 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

weizhoublue Aug 22, 2023 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

weizhoublue Aug 22, 2023 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

weizhoublue Aug 22, 2023 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

weizhoublue commented Aug 22, 2023

Choose a reason for hiding this comment

Choose a reason for hiding this comment

weizhoublue Aug 22, 2023 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

ii2day commented Aug 23, 2023

Choose a reason for hiding this comment

Choose a reason for hiding this comment

weizhoublue Aug 23, 2023 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

weizhoublue commented Aug 24, 2023

codecov bot commented Aug 11, 2023 •

edited

Loading

weizhoublue Aug 15, 2023 •

edited

Loading

weizhoublue Aug 15, 2023 •

edited

Loading

weizhoublue Aug 22, 2023 •

edited

Loading

weizhoublue Aug 22, 2023 •

edited

Loading

weizhoublue Aug 22, 2023 •

edited

Loading

weizhoublue Aug 22, 2023 •

edited

Loading

weizhoublue Aug 22, 2023 •

edited

Loading

weizhoublue Aug 23, 2023 •

edited

Loading