Nikhil

@itroot we are connecting to `postgres` database and then issuing this command `CREATE DATABASE test_42 WITH template = postgres;` Note that the template DB to use is `postgres` itself. Kinda...

The current logic sets the number of processed tuples for the INSERT based on the input that we are sending to the DNs. In case of `ON CONFLICT`, the tuple...

So, ``` ANALYZE disttable_repl; ``` the above always sets the `search_path` before sending the ANALYZE command to the datanodes. The further steps which send calls to DNs use the `deparse_func_call`...

I was trying to look at ways in which we could get two different connections above. It's possible that a call to `cache_syscache_invalidate` might invalidate the `connection1` entry causing the...

So my theory about different connections being used is correct! The attached PG log has a failed dist_api_calls test. [PostgreSQL log ubuntu-20.04 Debug 12.12.zip](https://github.com/timescale/timescaledb/files/9525246/PostgreSQL.log.ubuntu-20.04.Debug.12.12.zip) ``` cluster_super_user [12841] db_dist_api_calls_1 LOCATION: exec_simple_query,...

@boxhock can you please share the stack trace of the crashing backend? You can use the below to get a stack trace on ubuntu: https://wiki.postgresql.org/wiki/Getting_a_stack_trace_of_a_running_PostgreSQL_backend_on_Linux/BSD#On_Ubuntu

@bodypumper will it be possible for you to share the schema of the involved table[s] and the views and queries in text form here?

@bodypumper do you mean you got the same results as you got with 1.x earlier with 2.5 as well?

@akamensky are you sure that you are facing OOM due to this database behavior on your setup? This looks like standard OS behavior in terms of caching.

@akamensky will it be possible to share your schema/data and the update queries to help us reproduce this locally?