Lukáš Král comments

Results 13 comments of


                                            Lukáš Král

`cannot complete operation on a table with identity column` error during `create_distributed_table()` on 12.1

We have recreated one of the clusters with only one worker to test if the issue persists and we keep encountering it even in 1 coordinator + 1 worker setup....

`cannot complete operation on a table with identity column` error during `create_distributed_table()` on 12.1

Unfortunately `table2` is fairly complex and references multiple other tables. I will see if I can trigger this with some simple table. We have also encountered this same exact issue...

`cannot complete operation on a table with identity column` error during `create_distributed_table()` on 12.1

Okay, so we can trigger on our clusters with just this: ``` START TRANSACTION; CREATE TABLE public.table2 ( id int8 GENERATED ALWAYS AS IDENTITY NOT NULL, CONSTRAINT table2_pk PRIMARY KEY...

refinery_cli: ERROR: cannot insert multiple commands into a prepared statement

I can, it is the very first migration we have. Did some stepping through the code and it is correctly called with batch_execute, but it always fails with above mentioned...

refinery_cli: ERROR: cannot insert multiple commands into a prepared statement

I have created a simple migration which triggers this issue: ``` SET LOCAL citus.multi_shard_modify_mode TO 'sequential'; CREATE SCHEMA IF NOT EXISTS "public"; -- public.table1 CREATE TABLE public.table1 ( id int8...

refinery_cli: ERROR: cannot insert multiple commands into a prepared statement

Can't really try it against just PostgreSQL as the `create_reference_table` functions are part of the Citus extension. I just wanted to say that we aren't talking about pure PostgreSQL here...

Distributed deadlock after issuing a simple `UPDATE` statement

I can confirm that we are indeed updating single row with single statement. This is log from worker 1 containing "deadlock detected", which mentions this occuring on the worker 2...

Distributed deadlock after issuing a simple `UPDATE` statement

After disabling deadlock detection, we can observe that there are indeed several transactions that are completely stuck. All of them have the same exact `UPDATE` (with different values). Below is...

Distributed deadlock after issuing a simple `UPDATE` statement

Unfortunately I don't have more that could help with reproduction right now, as I am unable to reproduce this on an another cluster with the same schema either. But I...

Distributed deadlock after issuing a simple `UPDATE` statement

It would be possible to grant an access to the faulty cluster temporarily if it would help with investigation of this issue. Let me know if this is something worth...