Describe the bug
When we use docker compose, first the findingaid service comes up, then the web service comes up. However, there is a race condition. There is a "depends_on" statement that makes sure that web waits for findingaid, but if web comes up faster than Docker's embedded DNS can set up "findingaid" in the docker network, then the web container cannot resolve DNS and crashes with [emerg] host not found in upstream "findingaid". There is a fix for the exploreuk-web-app omeka service in ukl-ansible-playbooks which can be repurposed for development's nginx configuration.
This fix will need to be addressed in production as well. It is my current suspicion that this is happening in production, but that the restart policy is restarting the containers until the DNS resolves.
Expected behavior
The containers should start without issue.
Desktop (please complete the following information):
- OS: Windows/Linux (through WSL)
Describe the bug
When we use docker compose, first the findingaid service comes up, then the web service comes up. However, there is a race condition. There is a "depends_on" statement that makes sure that web waits for findingaid, but if web comes up faster than Docker's embedded DNS can set up "findingaid" in the docker network, then the web container cannot resolve DNS and crashes with
[emerg] host not found in upstream "findingaid". There is a fix for the exploreuk-web-appomekaservice in ukl-ansible-playbooks which can be repurposed for development's nginx configuration.This fix will need to be addressed in production as well. It is my current suspicion that this is happening in production, but that the restart policy is restarting the containers until the DNS resolves.
Expected behavior
The containers should start without issue.
Desktop (please complete the following information):