Discourse: PG::InternalError (ERROR: subplan "SubPlan 2" was not initialized CONTEXT: parallel worker )

Discourse Discourse System Administration
1

Looks like we will need to rebuild our (single) data container soon to get rid of this error in Discourse. Or, I will just build Discourse for scratch and restore the data and switch over when it is done to minimize down time.

PG::InternalError (ERROR:  subplan "SubPlan 2" was not initialized
CONTEXT:  parallel worker
)
app/models/private_message_topic_tracking_state.rb:29:in `report'
app/controllers/users_controller.rb:323:in `private_message_topic_tracking_state'
app/controllers/application_controller.rb:399:in `block in with_resolved_locale'
app/controllers/application_controller.rb:399:in `with_resolved_locale'
lib/middleware/omniauth_bypass_middleware.rb:71:in `call'
lib/content_security_policy/middleware.rb:12:in `call'
lib/middleware/anonymous_cache.rb:356:in `call'
config/initializers/100-quiet_logger.rb:23:in `call'
config/initializers/100-silence_logger.rb:31:in `call'
lib/middleware/enforce_hostname.rb:23:in `call'
lib/middleware/request_tracker.rb:187:in `call'

This error appears when using PMs / DMs (direct messages) in my account; but no one else has reported it.

2
3

Rebuilding this data container as Jay did in that topic will cause the site to be off line for maybe 30 minutes.

4

afael dos Santos SilvaFalcoteam

28d

That is indeed a PostgreSQL bug that was solved in latest minor version. Can you do a rebuild? That will update the PostgreSQL to latest minor.

Ok I saw this at the time but as Jay said it was one user, I thought we were not affected... :confused:

5

Guess I will rebuild the data container this weekend…

6 
community# docker exec -it pg13  bash -c 'psql --version'
psql (PostgreSQL) 13.1 (Debian 13.1-1.pgdg100+1)
7

Rebuilding our data container did not work as per the meta instructions (above); but I managed to improvise and 30 minutes later, it's running again and all is good.

community# docker exec -it pg13  bash -c 'psql --version'
psql (PostgreSQL) 13.4 (Debian 13.4-4.pgdg100+1)

As I understand meta from the past, they consider multi-containers solutions "not supported" so it is not a surprise that the instructions above did not work for us. I knew going into the data container rebuild that it would not go smoothly per the meta-talk on this bug.

The good news, is that bug is quashed, I assume, with the PG upgrade and new data container.

See Also:

https://www.postgresql.org/docs/release/13.4/

8

Note:

After this upgrade, although not required, performed:

VACUUM ANALYZE:
REINDEX SCHEMA CONCURRENTLY public;

.... just because I was already working on the DB and seems like a good housekeeping thing to do.

The site does seem a little faster after housekeeping; but that could all be in my imagination. I notice our cars and motorcycles always seem to run better after we wash them :slight_smile: