Re: BUG #18433: Logical replication timeout

Hi

> 3. This log was got for a table size of 39 GB but we have tables much bigger than this.
>
> Our workflow for the tables size 39 GB
>
> 1) Download schema from the source database instance
>
> 2) Deleted PK, FK, and Indexes from the table schema(we did it to increase data load, startup process)
> 3)Upload the schema to the destination DB.
> 4) configurated identity replication full at source DB for this table
> 5) Configured Logical replication between source and destination DB this table
> 6) During catchup on this table process we got the messages in log that you can see above.
>
> 7) We also tried create primary key for this table during catchup state but this process was blocked by logical replication worker so we had to cancel primary key creation process.
>
> I want to highlight that we used this workflow for PostgreSQL 10 and it worked.
>

I tried the steps you provided to reproduce the issue on Postgresql 13
but, I was not able to reproduce it. I tried it for tables with size
~120GB and also tried it separately for table with size ~40GB.

Steps I used:
1. Created a sourceDB and DestinationDB.
2.
Populated sourceDB with data.
"create table t1(c1 int PRIMARY KEY, c2 int);"
"insert into t1 values(generate_series(1,1500000000), 2)"
"update t1 set c1 = 0 where c1 = 1"
3.
Deleted Primary Key on sourceDB
"alter table t1 drop constraint t1_pkey"
4.
Created schema on destinationDB
"create table t1(c1 int, c2 int);"
5.
Configured Replica identity full on sourceDB
"ALTER TABLE t1 REPLICA IDENTITY FULL"
6.
Configured Logical Replication Setup
on sourceDB: "create publication pub1_new for all tables"
on destinationDB: ""create subscription test1_new connection
'dbname=postgres host=localhost port=5432' publication pub1_new;"

For me the initial catchup was successful and I did not receive any
issue in sourceDB logs or destinationDB logs.
Can you point out if you did anything different?

Also, For me the initial catchup succeeds even if I did not configure
replica identity full on sourceDB
Can you point out how your steps were different to get error related
to replica identity full?

Also, are you running any queries concurrently on sourceDB or destinationDB?

Thanks and Regards,
Shlok Kyal