search

2ndQuadrant / pglogical

Logical Replication extension for PostgreSQL 17, 16, 15, 14, 13, 12, 11, 10, 9.6, 9.5, 9.4 (Postgres), providing much faster replication than Slony, Bucardo or Londiste, as well as cross-version upgrades.
http://2ndquadrant.com/en/resources/pglogical/
Other
1.03k stars 154 forks source link

Segmentation fault when trying to replicate data #446

Open sethideepak opened 1 year ago

sethideepak commented 1 year ago

I'm currently experiencing a segmentation fault while attempting to replicate data using pglogical. The issue arises when converting two non-partitioned tables to a partitioned table during the replication process.

To address this, I've adjusted the configuration on the provider node, setting the parameter synchronize_data to false in order to skip schema structure validation:

SELECT pglogical.replication_set_add_all_tables('default', ARRAY['public'], synchronize_data := false);

Here are the details of the source and target environments:

Source:

PostgreSQL: 9.5.25 pglogical: 2.4.3 Target:

PostgreSQL: 15.5 pglogical: 2.4.3

Notably, I successfully tested the replication process on PostgreSQL 15.4 with pglogical 2.4.3 in a development environment, where it worked seamlessly.

However, upon attempting the same process on PostgreSQL 15.5 in a production environment, I encountered a replication failure for one of the tables. The specific error message is as follows:

Background worker "pglogical apply 428648:1802888491" (PID 23195) was terminated by signal 11: Segmentation fault.

I am currently investigating potential causes, considering factors such as compatibility issues, changes in schema handling, and any specific considerations related to PostgreSQL 15.5. I will proceed with debugging the segmentation fault and seek guidance from the relevant communities to resolve this issue effectively.

camilb commented 1 year ago

Having the same issue from 14.10 to 15.5 in AWS RDS. I had to completely delete the target instance. Testing again with 15.4

ahollmann commented 10 months ago

@camilb What was your outcome of your tests? We are having the same issues - crashes with AWS Aurora on 14.8.

camilb commented 10 months ago

@camilb What was your outcome of your tests? We are having the same issues - crashes with AWS Aurora on 14.8.

@ahollmann That was my mistake. One of the tables had a calculated column, which triggered a replication timeout in AWS RDS. If the column was omitted, the target db crashed and entered in a loop of recoveries. The solution was to remove the table from replication and dump/import it manually.

ahollmann commented 9 months ago

Thanks for your reply it was helping to fix our segfaults.