[Bucardo-general] Can stopped bucardo server increase overall size of db and its slave?

[Shubham Jhandei]

I have a question, will bucardo be able to increase overall size of db?

In dec’18, we migrated to RDS from EC2 postgres, using bucardo replication. But due to some reasons, we had to stop using that and put the task in backlog. We stopped the bucardo syncing, and left the system as it is. Now after 3 months after we stopped it, we are facing an high increase rate in db size as compared to duration before.

we suspect bucardo is responsible because the time it all started matches with the date we stopped bucardo.

We have following logs when run on db server:

Log.bucardo
(53367) [Tue Jan 22 16:00:32 2019] MCP End of cleanup_mcp. Sys time: Tue Jan 22 16:00:32 2019. Database time: 2019-01-22 16:00:32.391908+00
(53367) [Tue Jan 22 16:00:32 2019] MCP Exiting
*** Error in `Bucardo VAC.': double free or corruption (!prev): 0x0000000002997e50 ***


root at db:~# bucardo status;
Fri Apr  5 09:47:04 UTC 2019

Name                      State    Last good               Time            Last I/D    Last bad                Time
=========================+========+=======================+===============+===========+=======================+===============
sync_db1         | Good   | Jan 10, 2019 15:10:50 | 2034h 36m 14s | 0/0       | none                  |
 sync_db2      | Good   | Jan 22, 2019 16:00:20 | 1745h 46m 44s | 0/1       | none                  |
 sync_db3        | Good   | Jan 10, 2019 15:10:51 | 2034h 36m 13s | 0/0       | none                  |
 sync_db4           | Bad    | Jan 22, 2019 11:57:44 | 1749h 49m 20s | 24/24     | Jan 22, 2019 16:00:29 | 1745h 46m 34s
 sync_db5 | Bad    | none                  |               |           | Jan 22, 2019 16:00:31 | 1745h 46m 33s
 sync_db6       | Good   | Jan 22, 2019 16:00:30 | 1745h 46m 34s | 0/0       | none                  |
 sync_fb7 | Good   | Jan 22, 2019 16:00:29 | 1745h 46m 35s | 0/6       | none                  |
 sync_db8      | Good   | Jan 10, 2019 15:10:51 | 2034h 36m 13s | 0/0       | none                  |

> As you can see Last good is of Jan 22, which is the date we stopped it. There is an increase in every db, but the dbs db4 and db5, are increasing with higher speed.


root at db:~# bucardo list syncs
Sync "sync_db1"           Relgroup "relgroup_db1"          [Active]
  DB group "dbgroup_db1" dest_db1:target source_db1:source
Sync "sync_db2"       Relgroup "relgroup_db2"      [Active]
  DB group "dbgroup_db2" dest_db2:target source_db2:source
Sync "sync_db3"         Relgroup "relgroup_db3"        [Active]
  DB group "dbgroup_db3" dest_db3:target source_db3:source
Sync "sync_db4"            Relgroup "relgroup_db4"           [Active]
  DB group "dbgroup_db4" dest_db4:target source_db4:source
Sync "sync_db5"  Relgroup "relgroup_db5" [Active]
  DB group "dbgroup_db5" dest_db5:target source_db5:source
Sync "sync_db6"        Relgroup "relgroup_db6"       [Active]
  DB group "dbgroup_db6" dest_db6:target source_db6:source
Sync "sync_db7"  Relgroup "relgroup_db7" [Active]
  DB group "dbgroup_db7" dest_db7:target source_db7:source
Sync "sync_db8"       Relgroup "relgroup_db8"      [Active]
  DB group "dbgroup_db8" dest_db8:target source_db8:source

> This makes me believe that all syncs are active, but we are sure that bucardo is not pushing it anywhere.

Also there is a slave server to this database, on which there is no bucardo service running but size is similar to master, which put us in strong doubts about the reason of the cause.

So my questions here is:
1. Will the above scenario possible?
2. Is there any command, to see if there are any locally stored replication files? if yes, where are they
3. Can I remove the above files through some command, or manually?

PS: our dba is not here, so need help.

Thanks,
Shubham Jhandei
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.endcrypt.com/pipermail/bucardo-general/attachments/20190405/f65cd15f/attachment-0001.html>