Configuration is pasted below. We replaced the old ID’s with the new IDs, tried to restart the HTTPD web servers and saw no reflection in the portal. We then tried to use /opt/ood/ood-portal-generator/sbin/update_ood_portal to no avail. Is there another service we have to restart? Will it cause interruptions to users? We also restarted the browser web server using the help dropdown. Nothing seems to work and I can’t find anything relative in the documentation besides the config itself.
This should be all that you need to do and yes it will cause a disruption if the user has shell connections.
If nothing is happening when you restart, then I’d wonder if the YAML format is correct. I don’t believe this is being cached anywhere.
Not sure how it could continue to keep the old configuration if you are in fact bouncing the webserver. It seems to me that it should either pick up the new config if it’s correct, or drop the old config if it’s not properly formatted YAML.
If it keeps the old configuration, I guess I’d have to question if it’s infact bouncing at all or if the new config has actually been deployed.
We tracked this down to an extra YAML file in the directory. We had copied the original file and left it with a .bak / date-based extension, and Open OnDemand did not tolerate having two YAML files present. Removing the backup file resolved the issue.
Credit to a teammate for spotting it, but that was the root cause and fix.