Skip to main contentMerative SPM on Kubernetes

MustGather data for SPM

Collecting MustGather data for SPM on Kubernetes

Introduction

The information laid out in this section details the information needed when raising a support case. Note that not all information detailed is necessary when raising an issue on GitHub.

Merative SPM may request the Helm Charts and Dockerfiles developed by the engineering team as part of the support process.

Question

What information should I collect for issues with Merative Social Program Management (SPM) in a Kubernetes environment to assist Meravite Cúram Support with their investigations?

Answer

First, a comprehensive description of the problem along with answers to the following questions:

  • Can you recreate the issue and provide the steps for Merative Support to do same?
  • What is the impact of the issue; for instance, frequency of occurrence and pervasiveness?
  • Have you identified any workarounds or mitigations?
  • What investigation have you already performed?

Collecting MustGather data guidelines

Outlines below are must gather data guidelines that we recommend are followed.

  • Do not provide any unanonymized data or information.
  • Save all output to text files or files of appropriate formats. Include the command that created the output file.
  • Logs and other artifacts can only be obtained from a Kubernetes pod if the pod is running; the two exceptions are:
    • stdout is still available via kubectl logs (see stdout logs from pods for detail) even if the pod is in a non-running state such as completed
    • logs that are stored persistently are available over the lifetime of that storage medium
  • Regarding command examples that follow:
    • In commands with variable information, such as namespaces, these are represented in the example commands by angle brackets, which you must replace.
    • Commands where additional options are possible are indicated by an ellipses: …

General SPM MustGather

Beyond the answers to these questions, you need to collect general third party software information (for example, versions) and more specific diagnostic, log, and other information for the areas where you are seeing the issue. This general information is described in the sections that follow. For issues with the following third party software you should consult their external MustGather links as appropriate:

The following sections describe how to gather information specific to SPM running on Kubernetes and provide information relating to where logs and traces reside relating to the MustGather.

SPM and base third party software versions

In a SPM environment the Ant configreport target includes the following versions in its generated zip file:

  • SPM version - available in the Installerlogs/Installhistory.txt file within the resulting $SERVER_DIR/config_report.zip file
  • Inside the ConfigReporter*.log file of the config folder within the resulting $SERVER_DIR/config_report.zip file you will find the:
  • Java version
    • Application server version
    • Database version
    • Ant version

To generate the $SERVER_DIR/config_report.zip file run the Ant configreport target:

cd $SERVER_DIR
./build.sh configreport

Please provide the config_report.zip file.

Runtime systems

Operating systems and platforms

Include this information for all relevant platforms used for deployment, where you run the Azure CLI or Minikube.

On most systems the following command provides basic platform and operating system information:

uname -a

Kubernetes and related software versions

The following commands display the relevant versions for Docker, Helm and Kubernetes:

  docker version # Provides version information for the Docker client and server
  docker info    # Provides general runtime information about Docker


  helm version   # Provides Helm version information


  # Depending on your environment, obtain the appropriate Kubernetes version information:
  minikube version   # If applicable
  oc version         # If applicable (OpenShift)
  kubectl version    # Applies to Minikube, Kubernetes and OpenShift environments

Helm information

For deployments, gather information relevant or appropriate to your environment and the issue faced.

For Helm use the command appropriate to your Helm version to list the deployments in the environment, which provides the release and chart names needed in later Helm commands:

helm list -a --namespace <name_space>

Verify Helm Chart correctness

Use the helm lint command to verify the correctness of your Helm Charts (prior to installation), for example:

cd $SPM_HOME/helm-charts/spm
helm lint .

You should receive successful output from the helm lint command like this:

==> Linting .
Lint OK


1 chart(s) linted, no failures

If you encounter any errors, correct these, and recreate your issue.

Helm Chart rendering

Running the helm template command processes the chart and generates output similar to what would be generated during deployment; however, the helm install command (see the following examples) provides additional information; notably overrides provided via the install. So, where possible, the output from helm install is preferable to that of helm template.

helm template <release_name> <chart_name> ...

Check the output to verify that the expected values appear as intended.

Helm Chart installation

If you encounter errors when installing Helm Charts use the helm install command with the following options to generate additional information

  • --debug - checks the generated manifests of a release without installing the chart
  • --dry-run - produces output similar to helm template but additionally executes the secrets and verifies the objects

For example:

helm install <release_name> <chart_name> --debug --dry-run ...

Verify Helm Chart installation

The following command provides more information about what has been deployed or is running in Kubernetes.

The helm get all command provides information equivalent to helm install --debug --dry-run. For example:

helm get all <release-name> ...

Kubernetes deployment information

Enquiring a Kubernetes environment

If running in a Minikube environment use the following command to confirm its status:

minikube status

Use the following commands to get information about the running Kubernetes environment.

Display basic information about the cluster configuration:

kubectl config view --namespace <name_space> ...

Display basic information about the nodes and pods in the cluster:

kubectl get nodes -o wide --namespace <name_space> ...
kubectl get pods -o wide --namespace <name_space> ...

For each of the pods displayed by the kubectl get pods command use the pod name as input to the following command to get more detailed information about that pod, for example:

kubectl describe pod <pod_name> --namespace <name_space> ...

A convenient way to run the kubectl describe pod <pod_name> --namespace <name_space> ... command for all the pods in a namespace is as follows:

mkdir pod
pods=`kubectl get pods -o name --namespace <name_space>`
for p in $pods ; do kubectl describe $p --namespace <name_space> > $p.txt ; done

The results will be in the pod folder created using the kubectl command.

MustGather to collect logs and other artifacts

The following sections identify how to obtain logs and artifacts for specific areas of Kubernetes.

Externalizing logs and artifacts inside a Kubernetes pod

Externalizing logs and artifacts from a Kubernetes pod varies by the type of pod. The sections that follow identify locations for this information. The kubectl command reference gives several examples of how to externalize files from within a pod, such as:

kubectl cp <name_space>/<pod_name>:/logs /tmp/logs
kubectl exec -n <name_space> <pod_name> -- tar cf - /logs | tar xf - -C /tmp

Note: Both of the above kubectl command formats produce the following spurious, benign error which you can ignore:

tar: Removing leading `/' from member names

Minikube logs

Use the minikube logs command to obtain logs for the Minikube environment.

You can set the logging level when starting Minikube via the --v option of the minikube start command. For instance, and to conveniently save the output at the same time:

minikube start --alsologtostderr --v=3 2>&1 | tee /tmp/minikube-start.log

Noting that the log output at startup differs from the output from the runtime log content via minikube logs.

See Minikube documentation for more information.

stdout logs from pods

The kubectl logs provides the stdout logs for a specified Kubernetes pod; so the log data provided is specific to the type of pod or software running within it. For instance, if a pod is running WebSphere Liberty this command provides the equivalent of the contents of the WebSphere Liberty console.log file and for other pods will provide stdout log output specific to that pod’s context; for instance MQ, Cúram XML Server, or Cúram batch.

The command has various options, such as --follow, --tail, and --since to control and display the log output. See the kubectl help logs command for option details.

Third party software logs and artifacts

Beyond basic stdout logging from a pod there may be additional logging or artifacts needed for investigating an issue. These logs and artifacts will be available within the pod’s file system or, in the case of object storage, in the persistent storage location specified Cloud Object Storage.

Further, the specific logs, artifacts, and their locations vary depending on the software running within the pod.

The following sections identify these various locations.

WebSphere Liberty stdout logs

WebSphere Liberty pods are identified by the pod naming pattern beginning with: <helm_release_name>-apps- or <helm_release_name>-uawebapp-.

Thus, if your Helm release was named “purpleclown” you would find pod names like these from a kubectl get pods command:

  • purpleclown-apps-curam-consumer-7c578b9b59-t487n
  • purpleclown-apps-curam-producer-6794fc9b94-4srxr
  • purpleclown-apps-rest-producer-6858d55bd-x6z6s
  • purpleclown-apps-rest-consumer-5567ddc4cd-qpwp7
  • purpleclown-uawebapp-54b56d5866-2mnk5

where the number of pods may vary depending on the applications deployed and the number of replicas specified.

The default log location for WebSphere Liberty is the /logs folder of the pod’s file system and, for example, can be accessed with the following command

kubectl exec --namespace <name_space> -it <pod_name> -- ls -l /logs

which would produce output like:

-rw-r----- 1 default root 62334 Apr 14 06:01 messages.log
-rw-r--r-- 1 default root  2134 Mar 19 06:44 messages_20.04.14_06.00.47.0.log
-rw-r----- 1 default root   973 Apr 14 06:00 trace.log

When using object storage the WebSphere Liberty logs, JVM dumps and garbage collection logs, and Cúram JMX statistics are symbolically linked to the pod’s file system in /tmp/logs and available in your COS bucket.

WebSphere Liberty configuration information

WebSphere Liberty configuration information can be found in the pod’s file system in /liberty/wlp/usr/servers/defaultServer, which contains:

server.xml
jvm.options
server.env
bootstrap.properties
adc_conf/

The WebSphere Liberty server package command can be used to zip up the server; but, the logs folder, being stored separately, must be handled directly from the /logs folder.

MQ logs and artifacts

MQ pods are identified by the pod naming pattern beginning with: <helm_release_name>-mqserverapps-.

So, if your Helm release was named “purpleclown” you would find pod names like the following from a kubectl get pods command:

  • purpleclown-mqserver-curam-557d64c76-7sm8r
  • purpleclown-mqserver-rest-dbb994dc8-4p5cv

where the number of pods may vary depending on the SPM applications deployed.

The location of MQ logs is inside the MQ pods is as documented in the MQ IBM Documentation. You can interrogate the various log, error, and trace locations identified in the MQ IBM Documentation (and its MustGather) by accessing the MQ pod’s shell; for example:

kubectl exec --namespace <name_space> -it <mq_pod_name> -- /bin/bash

Cúram XML Server logs and artifacts

Cúram XML Server pods are identified by the pod naming pattern beginning with: <helm_release_name>-xmlserver-.

So, if your Helm release was named “purpleclown” you would find pods named like the following from a kubectl get pods command:

  • purpleclown-xmlserver-86b588c784-z76mz

where the number of pods may vary depending on the number of replicas specified.

Using the kubectl exec command as illustrated will place you into the default XML Server directory /opt/ibm/Curam/xmlserver:

kubectl exec --namespace <name_space> -it <xmlserver_pod_name> -- /bin/bash

The log file in /opt/ibm/Curam/xmlserver/tmp/xmlserver.log is the same as the stdout produced by the kubectl logs command.

For more information about XML server tracing and logging, see Configuring the XML server** in the Social Program Management XML Infrastructure Guide*. Gathering this data is beyond the scope of this document.

The Social Program Management PDF documentation is available to download from Merative Support Docs.

Cúram batch logs and artifacts

Cúram batch pods are identified by the pod naming pattern beginning with: <helm_release_name>-batch-.

So, if your Helm release was named “purpleclown” you would find pods named like the following from a kubectl get pods command:

  • purpleclown-batch-1587036600-9rrnw

where the number of pods may vary depending on the number of replicas specified.

Use the kubctl logs command to view the batch pod’s stdout.

HTTP Server logs and artifacts

The Apache HTTP Server is only used for hosting SPM static content this information would only be of interest for issues relating specifically to static content.

HTTP Server pods are identified by the pod naming pattern beginning with: <helm_release_name>-web-.

So, if your Helm release was named “purpleclown” you would find pods named like the following from a kubectl get pods command:

  • purpleclown-web-5c9cf5d868-ncdqk

where the number of pods may vary depending on the number of replicas specified.

For these pods the relevant log files can be found within the /var/log/httpd folder of the pod’s file system. For instance:

kubectl exec --namespace <name_space> -it <pod_name> -- ls -l /var/log/httpd/error_log /var/log/httpd/access_log

would show the relevant logs:

-rw-r--r-- 1 default root    7826 Jul 28 09:32 /var/log/httpd/access_log
-rw-r--r-- 1 default root    3297 Jul 28 09:25 /var/log/httpd/error_log
Previous
Troubleshooting: XML Server tracing
Next
Known issues