Select to view content in your preferred language

ArcGIS Enterprise for Kubernetes - EKS deployment issues

1083
5
08-03-2023 03:57 AM
JosephSia2
New Contributor III

I'm trying to set up and configure version 11.1 but have encountered issues setting up the organization. For some reason the process is failing when it tries to set up the dynamic persistent volumes. It does manage to set up the dynamic persistent volume for the configuration store however.

Would anyone have any pointers to what i should look for? It's rather confusing when it works for one but not for the others. Log output attached.

20230803 - Fail.png

5 Replies
JosephSia2
New Contributor III

Just an update that static provisioning of the persistent volumes doesn't make any difference. Am pretty much stuck and will likely need to reconsider using this. 

Am still getting the following final logged Warning after 5 minutes of running where it fails and I have no clue what the issue is.

2023-08-04T09:20:30,315|WARNING|35e76aaa-d56a-4ba9-97b3-992bfccf0c90||Enterprise Admin|1084|1|14|ElasticSearchManager|Managed spatiotemporal and index store is accessible but not ready to use yet. Timed out(60m).||

20230804 - Fail.png

0 Kudos
JosephSia2
New Contributor III

Spent the day plugging away at it. Turned on logging for EKS to try to understand what is occurring but to no avail.

JosephSia2_0-1691236423374.png

After poking around for a while, it dawned on me that even if I knew what was wrong, I am not in a position to rectify the issue. Nor do I know whom to seek to rectify it. It would likely not be rectified in a timely manner either.

The error is still a stubborn "Enterprise Admin|1084|1|14|ElasticSearchManager|Managed spatiotemporal and index store is accessible but not ready to use yet. Timed out(60m).||". And I am very certain it's not failing due to a timeout since the error indicates 60 minutes and the configuration process ran for less than 5 minutes.

JosephSia2_2-1691237680075.png

 

As expected, one of the statically provisioned Persistent Volumes has been bound for the config store. But it still doesn't give me a clue about what is wrong.

JosephSia2_1-1691236822271.png

Will look at the .install directory to try and see if there are further clues but I'm about ready to throw in the towel and revert to the better understood enterprise. 

0 Kudos
JosephSia2
New Contributor III

Tried tracing the .install directory. While I can somewhat make out what the commands are doing, I can't tell what the intent and grand design of the configuration process is, so am unable to make sense of it.  

0 Kudos
ChristopherPawlyszyn
Esri Contributor

Few questions here:

  • Do you see anything reflected in the cluster-level Events about provisioning issues?
  • Can you capture the relational store description and pod logs?
    • kubectl describe pod/arcgis-relational-store-<clusterID>-<nodeID>-0 -n <namespace>
    • kubectl logs pod/arcgis-relational-store-<clusterID>-<nodeID>-0 -n <namespace> -c main-container
  • Can you capture the Admin API pod logs to make sure nothing was missed in the log export?
    • kubectl logs pod/arcgis-rest-administrator-api-<ID>-<podID> -n <namespace> -c main-container

The spatiotemporal message is indeed a red herring, due to not being able to ship logs to the Opensearch endpoint from filebeat; we'll look at improving that message before the site is created.


-- Chris Pawlyszyn
0 Kudos
hgupta_22
New Contributor

does anyone has setup the ArcGIS Imagery platform on Kubernetes? I am trying to know potential issues that may occur with Kubernetes deployment and components that are not supported. Any insights would be greatly appreciated.

 

0 Kudos