If docker containers are re-started, we can't get the stopPlaces from tiamat unless we reload hasura metadata (there is already a ticket about this) from hasura console at http://localhost:3201/console. This reload fixes the problem that we can fetch the stop place information from tiamat. But there is another problem. After the reload, for some reason we can't mutate the stop places anymore, this error is thrown:
This is most likely caused by Tiamat losing track of what NeTEx id:s have already been used. They are stored in a table in the stop registry database, but also use Hazelcast to communicate with other Tiamat instances to make sure IDs are not reused. This causes Tiamat to not directly use the database to keep track of used IDs and rather uses it only on startup. If there was some incensistency when waking up Tiamat docker image, there is a possibility the IDs were not properly initialized.
Steps to reproduce:
Setup-dependencies-and-seed
Use UI (not yet in main) to mutate stop place (This should be successful)
Restart docker and wait for containers to be up
Reload metadata
Try to do another mutation via UI: Fails.
After this I used hasura console to do a simple mutation:
If docker containers are re-started, we can't get the stopPlaces from tiamat unless we reload hasura metadata (there is already a ticket about this) from hasura console at
http://localhost:3201/console
. This reload fixes the problem that we can fetch the stop place information from tiamat. But there is another problem. After the reload, for some reason we can't mutate the stop places anymore, this error is thrown:This is most likely caused by Tiamat losing track of what NeTEx id:s have already been used. They are stored in a table in the stop registry database, but also use Hazelcast to communicate with other Tiamat instances to make sure IDs are not reused. This causes Tiamat to not directly use the database to keep track of used IDs and rather uses it only on startup. If there was some incensistency when waking up Tiamat docker image, there is a possibility the IDs were not properly initialized.
Steps to reproduce:
After this I used hasura console to do a simple mutation:
Which was successful, and after that the UI worked again. Don't know why.