ibm/cp4waiops-samples

cp4waiops-samples's Issues

uninstall targets installations CRD instead of installations.orchestrator.aiops.ibm.com

The main uninstall script uses the installation CRD to find the AIOps instance, when in actuality the CRD is installations.orchestrator.aiops.ibm.com. At least this is the case on a ROKS install. The uninstall script fails b/c there are no installations.

move getting started to top

• move getting started to top
• link back to KC for install

@taylormgeorge91 @kvijai82

Uninstall hangs while deleting IAF PVCs; incomplete deletion of resources

Description

When running the uninstall script, it will hang indefinitely trying to delete data-iaf-system-elasticsearch-es-aiops-0.
The hanging issue is due to IAF PVC deletion occurring before deletion of IAF operators.

Additionally, it does not handle the deletion two resources under the ibm-common-services namespace:
nss-odlm-scope
odlm-scope-managedby-odlm
(Can be seen by running oc get namespacescopes.operator.ibm.com -n ibm-common-services)

Notes:
Found through automated testing pipeline; some tests may have been creating user resources on the PVC causing the operator pod to hold on to it, which is a scenario that may not have been previously tested.

Resolution

Fixes are made in this pull request. IAF PVC deletion is moved after IAF operators and the two resources are added to deletion list.

Initiate the folder structure for sample app work

As we start the work on sample app for CP4WAIOps, including documents and automation for people to quickly setup demo env or PoC of CP4WAIOps along w/ sample app, we need to define a folder structure to host all these contents. This issue is used to track the progress of the folder structure construction.

A proposed structure might look as below:

- sample-app (root folder for all sample app work)
  - data (sample data)
  - scripts (util scripts called by tasks)
  - tasks (ansible playbooks)
  - templates (j2 templates)
  sample-app.conf (configuration file)
  xyz.sh (shell script to start the automation)

Infrastructure Management Install Fails on MacOS due to incorrect usage of sleep

The installation scripts for infrastructure automation fails on the first run-through as the scripts use an 's' at the end of the sleep instruction to specify the length to sleep for. On linux this is supported but on MacOS only supports an integer after sleep in the shell. This causes the installation script to fail quickly as sleep returns instantly rather than waiting, meaning the script doesn't give the cluster time to deploy resources

To support MacOS sleep should not use an 's' after specifying the number of seconds, this will also support Linux as it default to seconds if no 's' 'm' or 'h' is put after the number.

Prereq checker tool for WAIOps 3.5 "uprates" errors (info as warning and warning as errors)

Examples:

[INFO] Checking if the job 'cp4waiops-entitlement-key-test-job' already exists.
Error from server (NotFound): jobs.batch "cp4waiops-entitlement-key-test-job" not found

That is not an error, it is just the involuntary output of a check for its existence. There is a --ignore-not-found=true flag for oc get that can suppress that error message.

[INFO] Checking storage providers
error: the server doesn't have a resource type "storagecluster"
error: the server doesn't have a resource type "storagecluster"

That is not an error, once again, just the involuntary output of a resource check. The cluster may just be using a different form of storage.

[WARNING] StorageCluster instance is not Online. In order for Portworx to work, an instance of StorageCluster must have a status of "Online".
No resources found in openshift-storage namespace.

StorageCluster's absence does not warrant warnings unless ODF is installed.

[WARNING] Openshift Data Foundation not running.

That is information, not reason to warn a user who is not attempting to use ODF.

Error from server (NotFound): storageclasses.storage.k8s.io "ibmc-file-gold-gid" not found

Same observation about involuntary output of oc get calls to check for resource existence. From the perspective of the user, this is not an Error.

[WARNING] ibmc-file-gold-gid is not running.

ibmc-file-gold-gid is a storage class, not something that runs. Also, it is perfectly fine for it to not exist on most clusters since ROKS is just one of many supported platforms.

[ERROR] At least one of the three Storage Providers are required
[ERROR] The supported Storage Providers are Portworx, Openshift Data Foundation, or IBMC File Gold Gid. See https://ibm.biz/storage_consideration_350 for details.

According to the cited page, these providers are "recommended" rather than "required". Therefore this should be a WARNING, not an Error.

Also, the cited page mentions the support for Spectrum Fusion, so that the final error message is incomplete.

4.4 prereq script does not comply with documentation.

Hello!

The 'prereq.sh' script at line 170 does not create test job inside cp4aiops namespace as pointed in IBM Cloud Pak for AIOps Documentation.

Manually adding entitlement key pull secret in default namespace allow 'prereq.sh' to run, but does not really check installation compliance due to this mismatch.

It's possible to fix either CloudPak documentation or this script?

Thanks!

check_additional_installation_exists needs updated with proper installation CRD

We fixed the problem with the incorrect CRD for deleting the installation but the check_additional_installation_exists method still targets installation instead of installations.orchestrator.aiops.ibm.com.

v3.3 uninstall hangs on delete_connections

We had an AIOps v3.3 install that didn't fully complete. We ran the v0.3 version of the uninstall script and for the delete_connections step it hung. There were no connections to delete, so it seems that this case is not properly handled by the script. To workaround we had to comment out the call to delete_connections and run again.