排除CNDP上的Kubernetes Pod崩溃故障

简介

本文档介绍如何对云本地部署平台(CNDP)上的Pod崩溃进行故障排除。

先决条件

要求

本文档没有任何特定的要求。

使用的组件

本文档不限于特定的软件和硬件版本。

本文档中的信息都是基于特定实验室环境中的设备编写的。本文档中使用的所有设备最初均采用原始（默认）配置。如果您的网络处于活动状态，请确保您了解所有命令的潜在影响。

背景信息

在此设置中，云本地部署平台(CNDP)托管会话管理功能(SMF)。

问题

您会看到有关Pod崩溃的通用执行环境(CEE)的警报。

Command:
cee# show alerts active summary summary

Example:
[smf-rcdn/cee-rcdn] cee# show alerts active summary summary
NAME UID SUMMARY
--------------------------------------------------------------------------------------------
k8s-pod-crashing-loop bd4394046466 Pod smf-rcdn/smf-service-n0-6 (smf-service) is...
k8s-pod-crashing-loop 0ac1019911e3 Pod smf-rcdn/smf-service-n0-14 (smf-service) i...
k8s-pod-crashing-loop eeff8fa16660 Pod smf-rcdn/smf-service-n0-9 (smf-service) is...
k8s-pod-crashing-loop 470ff66822dc Pod smf-rcdn/smf-service-n0-5 (smf-service) is...
k8s-pod-crashing-loop cc8950f07ace Pod smf-rcdn/smf-service-n0-15 (smf-service) i...
k8s-pod-crashing-loop 05a7d1e291a6 Pod smf-rcdn/smf-service-n0-3 (smf-service) is...

分析

连接到主节点并显示所有崩溃的Kubernetes Pod。Grep for CrashLoopBackOff.  从相同输出中，我们可以看到此Pod重新启动的次数。 

Command:
master$ kubectl get pods -n 
      
        |grep -v CrashLoopBackOff 
       
 
       
Example: 
      
cloud-user@smf-rcdn-master-1:~$ kubectl get pods -n smf-rcdn |grep -v Running
NAME                                                        READY   STATUS             RESTARTS   AGE
smf-service-n0-10                                           1/2     CrashLoopBackOff   1224       6d7h
smf-service-n0-11                                           1/2     CrashLoopBackOff   1242       6d7h
smf-service-n0-15                                           1/2     CrashLoopBackOff   1244       6d7h
smf-service-n0-2                                            1/2     CrashLoopBackOff   1241       6d7h
smf-service-n0-3                                            1/2     CrashLoopBackOff   1251       6d7h
smf-service-n0-5                                            1/2     CrashLoopBackOff   1231       6d7h
smf-service-n0-7                                            1/2     CrashLoopBackOff   1249       6d7h

描述崩溃的Pod。这样您就可以了解有关Pod崩溃原因的更多详细信息。在Events（事件）下观察日志。

Command:
master$ kubectl describe pod -n 
       
       
         |grep -i start 
        
 
        
Example 
        
      :
cloud-user@smf-rcdn-master-1:~$ kubectl describe pod -n smf-rcdn smf-service-n0-11 |grep -i start Start Time: Tue, 09 Aug 2022 03:13:54 +0000 Started: Tue, 09 Aug 2022 03:13:56 +0000 Restart Count: 0 Started: Mon, 15 Aug 2022 11:33:10 +0000 Started: Mon, 15 Aug 2022 11:26:55 +0000 Restart Count: 1263 Started: Tue, 09 Aug 2022 03:13:58 +0000 Restart Count: 0  Events:  Type Reason Age From Message ---- ------ ---- ---- ------- Warning BackOff 65s (x15210 over 3d6h) kubelet Back-off restarting failed container

例如，您有Pod smf-service-n1-0 发生故障时，您需要连接到 smf-rcdn-service-ims2 收集核心文件。

ubuntu@smf-rcdn-master1:~$ kubectl get pods -n smf-ims -o wide | grep smf-service-n1-0 
NAME                READY   STATUS    RESTARTS   AGE   IP          NODE                    NOMINDATEDN NODE   READINESS GATES
smf-service-n1-0    2/2     Running   10         9h    10.20.9.142 smf-rcdn-service-ims2         
     
      
       
       
     

连接到节点是崩溃并收集二进制文件的主机Pod。思科分析需要此文件。

Command:
master1:~$  kubectl cp 
      
        / 
       
         :/opt/workspace/smf-service /tmp/smf-service  
        
      

Example:
ubuntu@smf-rcdn-master1:~$ kubectl cp smf-ims/smf-service-n1-0:/opt/workspace/smf-service /tmp/smf-service 

连接到节点是崩溃的主机Pod，并转至文件夹/var/lib/systemd/coredump/并显示内容。如果已生成，您可以在此文件夹中看到它们。

Example:
ubuntu@smf-rcdn-master1:~$ ssh smf-rcdn-service-ims2
ubuntu@smf-rcdn-service-ims2:~$ cd /var/lib/systemd/coredump/
ubuntu@smf-rcdn-service-ims2:/var/lib/systemd/coredump$ ls -ltr
total 982340
-rw-r----- 1 root root 52968460 Sep 21 16:40 core.smf-service.0.a829fbabe2e649a7ab02150838fe47ae.1232.1599842408000000.lz4
-rw-r----- 1 root root 61609776 Sep 21 16:41 core.smf-service.0.a829fbabe2e649a7ab02150838fe47ae.3468.1599842463000000.lz4
-rw-r----- 1 root root 74233259 Sep 21 16:46 core.smf-service.0.a829fbabe2e649a7ab02150838fe47ae.28259.1599842775000000.lz4
-rw-r----- 1 root root 58241763 Sep 21 16:52 core.smf-service.0.a829fbabe2e649a7ab02150838fe47ae.17155.1599843174000000.lz4
-rw-r----- 1 root root 43732684 Sep 21 16:56 core.smf-service.0.a829fbabe2e649a7ab02150838fe47ae.3076.1599843385000000.lz4
-rw-r----- 1 root root 52377930 Sep 21 17:06 core.smf-service.0.a829fbabe2e649a7ab02150838fe47ae.8024.1599844002000000.lz4
-rw-r----- 1 root root 63990106 Sep 21 17:07 core.smf-service.0.a829fbabe2e649a7ab02150838fe47ae.26962.1599844074000000.lz4
-rw-r----- 1 root root 98058261 Sep 21 17:15 core.smf-service.0.a829fbabe2e649a7ab02150838fe47ae.13026.1599844546000000.lz4
-rw-r----- 1 root root 59586871 Sep 21 17:24 core.smf-service.0.a829fbabe2e649a7ab02150838fe47ae.21720.1599845052000000.lz4
-rw-r----- 1 root root 71187759 Sep 21 17:50 core.smf-service.0.a829fbabe2e649a7ab02150838fe47ae.19705.1599846648000000.lz4
-rw-r----- 1 root root 96949278 Sep 21 17:57 core.smf-service.0.a829fbabe2e649a7ab02150838fe47ae.11744.1599847049000000.lz4
-rw-r----- 1 root root 6052439 Sep 21 17:57 core.smf-service.0.a829fbabe2e649a7ab02150838fe47ae.23846.1599847052000000.lz4
-rw-r----- 1 root root 70642243 Sep 21 17:58 core.smf-service.0.a829fbabe2e649a7ab02150838fe47ae.18327.1599847110000000.lz4
-rw-r----- 1 root root 66052273 Sep 21 18:10 core.smf-service.0.a829fbabe2e649a7ab02150838fe47ae.1504.1599847843000000.lz4
-rw-r----- 1 root root 65132876 Sep 21 18:10 core.smf-service.0.a829fbabe2e649a7ab02150838fe47ae.12528.1599847855000000.lz4
-rw-r----- 1 root root 65000665 Sep 21 18:32 core.smf-service.0.a829fbabe2e649a7ab02150838fe47ae.9462.1599849167000000.lz4
ubuntu@smf-rcdn-master1:~$:/var/lib/systemd/coredump$  

压缩文件夹中的所有文件。

ubuntu@smf-rcdn-service-ims2:~$ sudo tar czvfsmf-rcdn-service-ims2.tar.gz *.lz4  

从主SFTP到核心所在的节点，并将它们下载到主/tmp文件夹，然后将其拉到PC。

ubuntu@smf-rcdn-master1:~$: sftp smf-rcdn-service-ims2

命令会在上次Pod重新启动之前打印日志，并捕获崩溃的签名。

Command:
master:~$ kubectl logs -n 
     
     
       -p 
      
        -c 
       
           
        
 
        
 
        Example: 
        
ubuntu@smf-rcdn-master1:~$ 
        kubectl logs -n smf-ims -p smf-service-n1-0 -c smf-service  /usr/local/go/src/runtime/asm_amd64.s:1357 (0x462d01) panic: runtime error: invalid memory address or nil pointer dereference [signal SIGSEGV: segmentation violation code=0x1 addr=0x50 pc=0x13d92f6] goroutine 839296 [running]: panic(0x196c320, 0x3441300) /usr/local/go/src/runtime/panic.go:722 +0x2c2 fp=0xc000a9d050 sp=0xc000a9cfc0 pc=0x432d82 runtime.panicmem(...) /usr/local/go/src/runtime/panic.go:199 runtime.sigpanic() /usr/local/go/src/runtime/signal_unix.go:394 +0x3ec fp=0xc000a9d080 sp=0xc000a9d050 pc=0x4487cc smf-service/userplane.(*UpfServData). 
        ProcessSessionModificationResponse(0xc0059fe660, 0xc005b98f00, 0xc00aa6e3c0, 0x2001181ae72b892, 0xc00ea43570, 0x3, 0x4, 
        
0xc005cd0820, 0xc005b11410, 0xc005b10d20, ...) /opt/workspace/smf-service/src/smf-service/userplane/upfSessionModification.go:743 +0x526 fp=0xc000a9d408 sp=0xc000a9d080 pc=0x13d92f6 smf-service/procedures/4g/pdn5g4gHo.(*Pdn5g4gHoProcedure).awtUpfModifyProcN4ModifyResp(0xc005a17440, 0xc0099e36c0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0) /opt/workspace/smf-service/src/smf-service/procedures/4g/pdn5g4gHo/mbrUtils.go:485 +0x24d fp=0xc000a9d630 sp=0xc000a9d408 pc=0x1562d0d smf-service/procedures/4g/pdn5g4gHo.(*Pdn5g4gHoProcedure).handleUpfModifyEvents(0xc005a17440, 0xc0099e36c0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0) /opt/workspace/smf-service/src/smf-service/procedures/4g/pdn5g4gHo/stateHandler.go:196 +0x4a1 fp=0xc000a9d768 sp=0xc000a9d630 pc=0x1570d31 smf-service/procedures/4g/pdn5g4gHo.(*Pdn5g4gHoProcedure).HandleEvent(0xc005a17440, 0xc0099e36c0, 0x6, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, ...) /opt/workspace/smf-service/src/smf-service/procedures/4g/pdn5g4gHo/procedure.go:364 +0x707 fp=0xc000a9d8d0 sp=0xc000a9d768 pc=0x1567887 smf-service/vendor/wwwin-github.cisco.com/mobile-cnat-smf/smf-common.git/src/smf-common/callflow.(*BaseProcedure).Handle(0xc00568b4a0, 0xc0099e36c0, 0x0, 
        
0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0) /opt/workspace/smf-service/src/smf-service/vendor/wwwin-github.cisco.com/mobile-cnat-smf/smf-common.git/src/smf-common/callflow/BaseProcedure.go:54 +0xdb 
        
fp=0xc000a9d978 sp=0xc000a9d8d0 pc=0xf5996b smf-service/vendor/wwwin-github.cisco.com/mobile-cnat-smf/smf-common.git/src/smf-common/callflow.(*SessionState).ProcessContinue(0xc00b79b6d0, 0xc0099e36c0, 
        
0xc00568b4a0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, ...) /opt/workspace/smf-service/src/smf-service/vendor/wwwin-github.cisco.com/mobile-cnat-smf/smf-common.git/src/smf-common/callflow/SessionState.go:169 +0x1f2 
        
fp=0xc000a9da20 sp=0xc000a9d978 pc=0xf5d552 smf-service/processor.(*SmfAppMessageProcessor).ProcessContinue(0x3a31da0, 0xc005b98f00, 0x1d34988, 0x35, 0x9, 0x1d34988, 0x35) /opt/workspace/smf-service/src/smf-service/processor/grpc_message_processor.go:430 +0x4ab fp=0xc000a9dc20 sp=0xc000a9da20 pc=0x174fc0b smf-service/vendor/wwwin-github.cisco.com/mobile-cnat-golang-lib/app-infra.git/src/app-infra/infra.(*masterBlueprint).processTransaction 
        
(0xc0003141e0, 0xc005b98f00, 0xc000a9dd98) /opt/workspace/smf-service/src/smf-service/vendor/wwwin-github.cisco.com/mobile-cnat-golang-lib/app-infra.git/src/app-infra/infra/MasterBlueprint.go:301 
        
+0x1a7 fp=0xc000a9dce8 sp=0xc000a9dc20 pc=0xd39ca7 smf-service/vendor/wwwin-github.cisco.com/mobile-cnat-golang-lib/app-infra.git/src/app-infra/infra.(*masterBlueprint). 
        
processTransactionWithCR(0xc0003141e0, 0xc005b98f00, 0x1cfeb00) /opt/workspace/smf-service/src/smf-service/vendor/wwwin-github.cisco.com/mobile-cnat-golang-lib/app-infra.git/src/app-infra/infra/MasterBlueprint.go:234 
        
 +0x394 fp=0xc000a9de78 sp=0xc000a9dce8 pc=0xd396e4 smf-service/vendor/wwwin-github.cisco.com/mobile-cnat-golang-lib/app-infra.git/src/app-infra/infra.(*masterBlueprint). 
        
processSessionTransaction(0xc0003141e0, 0xc005b98f00, 0x1, 0x0) /opt/workspace/smf-service/src/smf-service/vendor/wwwin-github.cisco.com/mobile-cnat-golang-lib/app-infra.git/src/app-infra/infra/MasterBlueprint.go:177 
        
+0x124 fp=0xc000a9ded0 sp=0xc000a9de78 pc=0xd39104 smf-service/vendor/wwwin-github.cisco.com/mobile-cnat-golang-lib/app-infra.git/src/app-infra/infra.(*masterBlueprint). 
        
processEvent(0xc0003141e0, 0xc005b98f00, 0x1d02487) /opt/workspace/smf-service/src/smf-service/vendor/wwwin-github.cisco.com/mobile-cnat-golang-lib/app-infra.git/src/app-infra/infra/MasterBlueprint.go:138 +0x5fc 
        
 fp=0xc000a9df88 sp=0xc000a9ded0 pc=0xd3869c smf-service/vendor/wwwin-github.cisco.com/mobile-cnat-golang-lib/app-infra.git/src/app-infra/infra.(*ApplicationContext).NewTransaction.func2 
        
(0xc0006af400, 0xc005b98f00) /opt/workspace/smf-service/src/smf-service/vendor/wwwin-github.cisco.com/mobile-cnat-golang-lib/app-infra.git/src/app-infra/infra/ApplicationContext.go:1268 
        
+0x7c fp=0xc000a9dfd0 sp=0xc000a9df88 pc=0xd9b69c runtime.goexit() /usr/local/go/src/runtime/asm_amd64.s:1357 +0x1 fp=0xc000a9dfd8 sp=0xc000a9dfd0 pc=0x462d01 created by smf-service/vendor/wwwin-github.cisco.com/mobile-cnat-golang-lib/app-infra.git/src/app-infra/infra.(*ApplicationContext).NewTransaction /opt/workspace/smf-service/src/smf-service/vendor/wwwin-github.cisco.com/mobile-cnat-golang-lib/app-infra.git/src/app-infra/infra/ApplicationContext.go:1266 +0x62c goroutine 1 [sleep]: runtime.gopark(0x1dbaa10, 0x34ef580, 0xc001f01313, 0x2) /usr/local/go/src/runtime/proc.go:304 +0xe0 fp=0xc000a3bca8 sp=0xc000a3bc88 pc=0x434ea0 runtime.goparkunlock(...)  
        
       
     

在出现Pod崩溃之前和之后连接到CEE并收集tac-debug。

tac-debug-pkg create from yyyy-mm-dd_hh:mm:ss to yyyy-mm-dd_hh:mm:ss tac-debug-pkg create from yyyy-mm-dd_hh:mm:ss to yyyy-mm-dd_hh:mm:ss 

行动计划

打开思科TAC的服务请求以查找此崩溃的根本原因。