incompatible change when switching from bitnami to OSI postgresql #212
Comments
We have created an issue in Pivotal Tracker to manage this: https://www.pivotaltracker.com/story/show/173076090 The labels on this github issue will be updated when the story is started. |
For visibility, @davewalter and I did a quick investigation and found:
|
On further investigation, we found that there appears to be an incompatibility between the bitnami image and the "official" docker image. Specifically, the bitnami image uses separate directories for the Postgres configuration (stored in As far as we can tell, the decision to use a different directory for the configuration file is baked into the bitnami image startup script, and is therefore not something we can override via the chart. |
We have reverted this commit for now. That should resolve this issue We have this story to investigate how to handle the postgres issue longer term. CC @davewalter |
if this is something team is interested in fixing, one solution might be to add an initContainer that moves data into expected locations (though be careful about making it idempotent). |
Closing this issue since we fixed the issue. |
Summary
This commit 44e0688 switched the
cf-db
postgresql image frombitnami/postgresql
torelintdockerhubpushbot/cf-for-k8s-postgres
. This corrupts the data on the volume when upgrading.Reproduction Steps
Can be reproduced in isolation
Starting the new image will print
After the new image has been started (and crashed) once you can also no longer start the old image:
starting and stopping the old image or new image multiple times is no problem.
See also
https://cloudfoundry.slack.com/archives/CH9LF6V1P/p1590670391438200
The text was updated successfully, but these errors were encountered: