EGIS cost savings - can we use just the EKS version of ArcGIS enterprise for testing in dev/ti without ill effect?

[EdisonOrellana-NOAA]

Investigate

Can we safely retire ArcGIS enterprise in dev/ti in favor of EKS ArcGIS as a cost savings measure?

Cost

• Do we need both versions of ArcGIS Enterprise in the Dev account to support our testing of pipelines and services or could we just use the EKS version even though in UAT and Prod we'd still have the regular version of ArcGIS Enterprise?
• EGIS tags in the Dev account spans the Dev and TI VPCs show that EGIS costs are ~$500 per day
• EKS: Not sure if there is a representation for the control nodes and service itself that we can tag

[shawncrawley]

I think the plan to test this and move forward with retiring the non-EKS version is pretty straightforward. It is this:

1. Refactor Terraform code to replace hard-coded if environment == "prod" ? "maps.water.noaa.gov" : ... logic with a variable passed in from the terraform environment variables file (i.e. these).
2. Refactor this line to hard-code "server" if the GIS_HOST environment variable contains "-testing" and to read it from the service_metadata otherwise.

I think that's about it. Once a PR is submitted with these updates, I can deploy to TI for testing. If it works, then we should be good to move forward with the plan to decommission the EC2-based EGIS machines.

[DrixTabligan-NOAA]

I agree. Might be a good idea to jump on the weekly call later with ESRI if you guys have any questions on how to publish and anything else specific related to the EKS deployment. Let me know and I'll add you to the meeting at 3PM ETC later. Drix Tabligan GAMA-1 Technologies SDB | ISO 20000 | ISO 9001 | ISO 27001 Operations  |  +1 (443) 538-7058 ***@***.***  |  www.gama1tech.com 7500 Greenway Center Drive, STE 400, Greenbelt, MD 20770 IMPORTANT: The contents of this email and any attachments are confidential. They are intended for the named recipient(s) only. If you have received this email by mistake, please notify the sender immediately and do not disclose the contents to anyone or make copies thereof. Join our Team!

…

On Nov 20, 2024 at 7:47 AM -0500, ShawnCrawley-NOAA ***@***.***>, wrote: I think the plan to test this and move forward with retiring the non-EKS version is pretty straightforward. It is this: 1. Refactor Terraform code to replace hard-coded if environment == "prod" ? "maps.water.noaa.gov" : ... logic with a variable passed in from the terraform environment variables file (i.e. these). 2. Refactor this line to hard-code "server" if the GIS_HOST environment variable contains "-testing" and to read it from the service_metadata otherwise. I think that's about it. Once a PR is submitted with these updates, I can deploy to TI for testing. If it works, then we should be good to move forward with the plan to decommission the EC2-based EGIS machines. — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you were assigned.Message ID: ***@***.***>

[shawncrawley]

After doing some testing today, I realized that to get EKS to work and be able to shut off the EC2-based system, we'll have to get the "MapxToSD" Geoprocessing Service published on the EKS server... @EdisonOrellana-NOAA

[EdisonOrellana-NOAA]

Tested in ti and has successfully published services and notified services for which data is missing or outdated. Done!

[EdisonOrellana-NOAA]

Reopening to keep ticket open for issues with service publish geoprocessing service.

• Bug only appeared later on after initial testing passed
• "subprocess" library is not available in EKS. This was used to run aws cli commands.
• Workaround is currently being implemented to use aws-c-s3 instead of the python subprocess library
  https://www.esri.com/content/dam/esrisites/en-us/media/manuals/arcgis-notebook-runtime/python-library-10.0.pdf