Hello Slony1,

we use the binary package of Slony 2.2.6-1 (32bit) with Windows10 
(32bit) together with Postgres 10.5-2 on the same system. Both installed 
with EnterpriseDB Stackbuilder.

I created a simple master and two slaves (both subscribing directly to 
the master) cluster. There no access from clients to all three databases 
at the moment. Experimental setting with 3 VBoxes.

But I have a question and run in two problems with this configuration.

1.: When I create my cluster, slonik logs a fallback to default 
Postgres8.4.mode. I know, that there is no support for postgres 10, but 
why doesnt slony fallback to the acual supported version of postgres?

2.: On several actions (switchover from master to one slave, 
resubscribing slave2 from master to slave1) the slon daemon simply stops 
running, without any failure logging. If I restart it, it somtimes runs 
on, sometimes it stops after some logging again.
Is this normal behavior?

3.: When I do a switchover from master to slave1, old master is still in 
sync with new master (formerly slave1). But slave2 is out of sync. I 
thought, with this switchover I should get a syncchain new master -> old 
master -> slave2.
I also tried to resubscribe slave2 to new master, but it didn't help.
Did I understand something wrong? What could be the reason of being out 
of sync with slave2?

What information should I post, to help you finding the issues (or my 
mistakes?).

Thank you, Wolle



-- 
May the soure be with you