Multiple modification to avoid DB corruption by RomainMou · Pull Request #82 · bimdata/quickstart-onpremise

RomainMou · 2026-01-28T09:57:26Z

Random corruption occurs over time on indexes.
To avoid this, we rework the upgrade process and
use a version of postgres where the libc doesn't change

Random corruption occurs over time on indexes. To avoid this, we rework the upgrade process and use a version of postgres where the libc doesn't change

Copilot

Pull request overview

Reworks parts of the Postgres upgrade/post-run process to reduce index corruption risk over time, and pins the Postgres Docker image tag to an OS-specific variant to avoid libc changes.

Changes:

Update default docker_postgres_tag to an OS-specific tag ({{ db_pg_version }}-trixie).
Add a new post-run upgrade step (20260128) intended to reindex/vacuum and refresh collation versions.
Enhance post-major-upgrade DB maintenance tasks (extensions update, reindex, collation refresh, vacuum in stages).

Reviewed changes

Copilot reviewed 5 out of 5 changed files in this pull request and generated 7 comments.

Show a summary per file

File	Description
roles/prepare_vars/defaults/main/docker_images.yml	Changes Postgres image tag default to include OS suffix.
roles/post_run/tasks/main.yml	Adds inclusion of the new 20260128 post-run DB maintenance tasks.
roles/post_run/tasks/db_upgraded.yml	Adds extension upgrade + reindex/collation refresh; changes vacuum behavior.
roles/post_run/tasks/20260128.yml	Introduces a new maintenance task file for the 20260128 upgrade step.
README.md	Updates documented default for `docker_postgres_tag`.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2026-01-28T10:02:17Z

roles/post_run/tasks/db_upgraded.yml

+        psql -U {{ lookup('vars', 'db_'+_db.name+'_user') }} -d "{{ _db.name }}" -c "
+          SELECT 'ALTER EXTENSION ' || quote_ident(extname) || ' UPDATE;'
+          FROM pg_extension;
+        "


This task currently only SELECTs ALTER EXTENSION ... UPDATE; statements and prints them; it does not actually execute the ALTER EXTENSION commands, so extensions will not be upgraded as intended. Use a DO block/loop (or generate commands and execute them) so the updates are applied.

Suggested change

psql -U {{ lookup('vars', 'db_'+_db.name+'_user') }} -d "{{ _db.name }}" -c "

SELECT 'ALTER EXTENSION ' || quote_ident(extname) || ' UPDATE;'

FROM pg_extension;

"

psql -U {{ lookup('vars', 'db_'+_db.name+'_user') }} -d "{{ _db.name }}" -v ON_ERROR_STOP=1 -c "DO \$\$DECLARE r record; BEGIN FOR r IN SELECT extname FROM pg_extension LOOP EXECUTE format('ALTER EXTENSION %I UPDATE', r.extname); END LOOP; END\$\$;"

Copilot · 2026-01-28T10:02:18Z

roles/post_run/tasks/20260128.yml

+# We should use docker_compose_v2_exec when ansible will be updated to 11 or another version that support it
+# With this module we could avoid relaying on container name


This file is missing the YAML document start marker (---). The repo's .yamllint enforces document-start: present: true, so this will fail linting.

Copilot · 2026-01-28T10:02:18Z

roles/post_run/tasks/20260128.yml

+- name: "Re-index the database to avoid corruption of some indexes on the database {{ _db.name }}"
+  community.docker.docker_container_exec:
+    container: "{{ db_info.name }}-db"


This task name interpolates {{ _db.name }}, but _db is not defined in this include (the loop var is db_info). Undefined variables in task names will raise errors in Ansible; update the task name to use db_info.name (and keep naming consistent across the file).

Copilot · 2026-01-28T10:02:18Z

roles/post_run/tasks/20260128.yml

+- name: "Run Postgres vaccum analyze on the database {{ db_info.name }} on the database {{ _db.name }}"
+  community.docker.docker_container_exec:


This task name references {{ _db.name }} (undefined here) and also contains duplicated text ("on the database ... on the database ..."). Update the name to use db_info.name and remove the duplication to avoid Ansible undefined-variable failures and improve readability.

Copilot · 2026-01-28T10:02:18Z

README.md

 | docker_postgres_image                   | "postgres"                                                       | Postgres docker image (use Dockerhub by default).                         |
-| docker_postgres_tag                     | "{{ db_pg_version }}-alpine"                                     | Postgres docker tag.                                                      |
+| docker_postgres_tag                     | "{{ db_pg_version }}-trixie"                                     | Postgres docker tag.                                                      |
 | docker_bimdata_tag                      | "{{ bimdata_version }}                                           | Docker tag use by all bimdata images.                                     |


The README table row for docker_bimdata_tag has a malformed default value (missing closing }}"), which breaks the markdown table formatting and shows an incorrect value to users. Please fix it to "{{ bimdata_version }}".

Suggested change

| docker_bimdata_tag | "{{ bimdata_version }} | Docker tag use by all bimdata images. |

| docker_bimdata_tag | "{{ bimdata_version }}" | Docker tag use by all bimdata images. |

Copilot · 2026-01-28T10:02:19Z

roles/prepare_vars/defaults/main/docker_images.yml


 docker_postgres_image: postgres  # Only use when use_external_db: false
-docker_postgres_tag: "{{ db_pg_version }}"
+docker_postgres_tag: "{{ db_pg_version }}-trixie" # use a tag with OS in it to avoid unplanified libc upgrade


The comment contains a spelling mistake: "unplanified" should be "unplanned" (and consider rephrasing to "avoid an unplanned libc upgrade").

Suggested change

docker_postgres_tag: "{{ db_pg_version }}-trixie" # use a tag with OS in it to avoid unplanified libc upgrade

docker_postgres_tag: "{{ db_pg_version }}-trixie" # use a tag with OS in it to avoid an unplanned libc upgrade

Copilot · 2026-01-28T10:02:19Z

roles/post_run/tasks/main.yml

+        - inventory_hostname in groups["db"]
+        - not use_external_db | bool
+        - db_info.current_version != ""
+        - db_info.current_version is version(db_pg_version, "==")


This include condition compares db_info.current_version (read from PG_VERSION and typically just the major version) to db_pg_version without normalizing db_pg_version the way other upgrade logic does (e.g., splitting on '.' to get the major). If db_pg_version is ever configured as a full semver (e.g. "17.2"), this condition will evaluate false and the 20260128 tasks will never run.

Suggested change

- db_info.current_version is version(db_pg_version, "==")

- db_info.current_version is version(db_pg_version.split('.') | first, "==")

Multiple modification to avoid DB corruption

4636ee4

Random corruption occurs over time on indexes. To avoid this, we rework the upgrade process and use a version of postgres where the libc doesn't change

RomainMou requested a review from Copilot January 28, 2026 09:57

Copilot started reviewing on behalf of RomainMou January 28, 2026 09:57 View session

Copilot AI reviewed Jan 28, 2026

View reviewed changes

Amoki approved these changes Jan 28, 2026

View reviewed changes

Multiples fixes.

1ac682a

Amoki approved these changes Jan 28, 2026

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Multiple modification to avoid DB corruption#82

Multiple modification to avoid DB corruption#82
RomainMou wants to merge 2 commits intomainfrom
fix/avoid-db-corruption-over-time

RomainMou commented Jan 28, 2026

Uh oh!

Copilot AI left a comment

Uh oh!

Copilot AI Jan 28, 2026

Uh oh!

Copilot AI Jan 28, 2026

Uh oh!

Copilot AI Jan 28, 2026

Uh oh!

Copilot AI Jan 28, 2026

Uh oh!

Copilot AI Jan 28, 2026

Uh oh!

Copilot AI Jan 28, 2026

Uh oh!

Copilot AI Jan 28, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

		# We should use docker_compose_v2_exec when ansible will be updated to 11 or another version that support it
		# With this module we could avoid relaying on container name

		- name: "Run Postgres vaccum analyze on the database {{ db_info.name }} on the database {{ _db.name }}"
		community.docker.docker_container_exec:

	\| docker_bimdata_tag \| "{{ bimdata_version }} \| Docker tag use by all bimdata images. \|
	\| docker_bimdata_tag \| "{{ bimdata_version }}" \| Docker tag use by all bimdata images. \|

	docker_postgres_tag: "{{ db_pg_version }}-trixie" # use a tag with OS in it to avoid unplanified libc upgrade
	docker_postgres_tag: "{{ db_pg_version }}-trixie" # use a tag with OS in it to avoid an unplanned libc upgrade

	- db_info.current_version is version(db_pg_version, "==")
	- db_info.current_version is version(db_pg_version.split('.') \| first, "==")

Conversation

RomainMou commented Jan 28, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Copilot AI Jan 28, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 28, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 28, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 28, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 28, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 28, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 28, 2026

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants