Closed bearpaws closed 1 month ago
Update: after increasing max_locks_per_transaction
to 10000, the ALTER EXTENSION completed successfully.
So I guess the question now is whether this is expected when dealing with a large number of tables and chunks, or is there a bug in the extension upgrade?
Generally speaking, what is the transaction strategy for an extension update? Are all table mutations made in a single transaction?
@bearpaws this is expected due to the numbers of relations you have. Updating the extension will require some locks on objects that depend of the extension and it will include hypertables and chunks. Default postgres configuration have a conservative value for max_locks_per_transaction
so in your case is recommended to increase this number. Even a simple pg_dump
will require a high value for this parameter.
What type of bug is this?
Unexpected error
What subsystems and features are affected?
Platform/OS
What happened?
Upgrading a timescale database from 2.11.2 (pg 14.9) to 2.15.2 (pg 14.12). When updating the timescaledb extension this error occurs:
After re-running it fails on the same
_timescaledb_internal._compressed_hypertable_1975
. But we increased themax_locks_per_transaction
setting and then it fails on a different table. So don't think it's related to any specific chunk.Increasing debug logging didn't offer anything more. Increased available memory too but same problem.
TimescaleDB version affected
2.15.2
PostgreSQL version used
14.12
What operating system did you use?
ubuntu docker images in kubernetes
What installation method did you use?
Docker
What platform did you run on?
On prem/Self-hosted
Relevant log output and stack trace
No response
How can we reproduce the bug?