Alembic Spanner DB Table Creation, Duplication Issue. #1481
-
|
While running Alembic migration file for GCP Spanner for creating the Table. The migration file runs and the table is created but the issue is that the migration tries to run again and as the required table has already been created it gives an Error that the the required table already exists in the schema and thus the migration fails. |
Beta Was this translation helpful? Give feedback.
Replies: 3 comments 17 replies
-
|
Hi, Is the alembic version table correctly updated to register that the migration was run? |
Beta Was this translation helpful? Give feedback.
-
|
No as migration is not successful and complete itself, there's only alembic_version table is created but it doesn't have any rows. |
Beta Was this translation helpful? Give feedback.
-
|
Sorry but I don't follow. You said
I interpreted this as the first run on "alembic upgrade" works fine, but trying to run it again causes the problem. |
Beta Was this translation helpful? Give feedback.
-
|
So, it goes as below: |
Beta Was this translation helpful? Give feedback.
-
|
that's very wired. alembic has no logic to re-run a single migration, so if there is no table on step iii and there is only a single create_table line in the migration file I don't know how that line could be run twice. Can you turn on the engine echo so that all the statement are logged to the console? |
Beta Was this translation helpful? Give feedback.
-
|
Yes, in step iii) there is no pre existing table in the DB |
Beta Was this translation helpful? Give feedback.
-
|
Hi, We are experiencing the same issue (except that we are trying to add a new column).
@kumarlti Did you find a solution for this problem? I am also pasting the full log here for reference. Thanks, Full log: |
Beta Was this translation helpful? Give feedback.
-
|
@CaselIT indeed, it's very strange... My last hope was a version update, so now I have tried with the following: I have the same results. Do you have any idea what could I try next? |
Beta Was this translation helpful? Give feedback.
-
|
I don't think that alembic is running the same migration again, at least not in the same What's strange is that at the end of the migration, when the connection is returned to the pool an error is raised by the driver. It seems the commit call does not wait for the server confirmation that takes a while to arrive and the exception is raised later. What would be expected is that either the single alter raise the error or at the very least when commit is issued the error is raised |
Beta Was this translation helpful? Give feedback.
-
|
I finally solved it! Now the log looks like this: Followup question: |
Beta Was this translation helpful? Give feedback.
-
|
You should be able to change the isolation level in the env in the connection: https://docs.sqlalchemy.org/en/20/orm/session_transaction.html#setting-transaction-isolation-levels-dbapi-autocommit |
Beta Was this translation helpful? Give feedback.
-
|
Thank you! |
Beta Was this translation helpful? Give feedback.
-
|
Adding a new comment so it can be marked as the answer for future reference. or connection/session level: |
Beta Was this translation helpful? Give feedback.
-
|
sometimes db require autocommit to perform some operation, for example historically postgresql required it when adding values to enums |
Beta Was this translation helpful? Give feedback.
Adding a new comment so it can be marked as the answer for future reference.
The solution to this probem was using
AUTOCOMMITmode that can be either set on block level:or connection/session level:
https://docs.sqlalchemy.org/en/20/orm/session_transaction.html#setting-transaction-isolation-levels-dbapi-autocommit