we are currently encountering an issue where running Alembic migrations (python -m alembic upgrade head) leads to build failures. the process exits with a "double free or corruption" error, which points towards a memory management issue.
double free or corruption (out)
release.sh: line 14: 326 Aborted python -m alembic upgrade head
INFO Main child exited normally with code: 134
INFO Starting clean up.
WARN hallpass exited, pid: 306, status: signal: 15 (SIGTERM)
2023/12/06 01:59:01 listening on [XXXXXX]:22 (DNS: [fdaa::3]:53)
[ 6.276142] reboot: Restarting system
machine restart policy set to 'no', not restarting
Error: release command failed - aborting deployment. error release_command machine YYYYYYYY exited with non-zero status of 134
this error does not occur when running simple Python commands, indicating the problem is specific to alembic or its interaction with our environment.
as a temporary measure, i have disabled alembic migrations to prevent build failures. this is not a sustainable solution as it bypasses important database schema updates. i'm opening this issue to keep a paper trail and ensure this temporary measure is well documented
it turns the culprit was the psycopg2 dependency which was causing memory corruption. installing just the binary https://pypi.org/project/psycopg2-binary/ fixes the issue...
we are currently encountering an issue where running Alembic migrations (
python -m alembic upgrade head
) leads to build failures. the process exits with a "double free or corruption" error, which points towards a memory management issue.this error does not occur when running simple Python commands, indicating the problem is specific to alembic or its interaction with our environment.
as a temporary measure, i have disabled alembic migrations to prevent build failures. this is not a sustainable solution as it bypasses important database schema updates. i'm opening this issue to keep a paper trail and ensure this temporary measure is well documented