while evaluating pgpool-II (replication mode) I ran into a situation
where I don't know the reason why....
The setup looks like:
host "one": postgres8.3.5 (defined as backend_0) and pgpoolII (port 9999)
host "two": postgres8.3.5 (defined as backend_1)
According to the tutorial I created database "bench_replication" and
inserted dummy data => everything's fine
In the next step I wanted to create some error scenarios:
a) shutdown "two" while a running update statement
aa) psql -p 9999 bench_replication
ab) update history set tid=<whatever>;
ac) while the update is running, kill postgres on host "two"
=> pgpool got noticed about that and cancelled the update
b) shutdown "one" while statement is running
ba) psql -p 9999 bench_replication
bb) update history set tid=<whatever>;
bc) kill postgres on host "one" (same host where pgpool is running)
=> nothing happens, psql never ends the statement, even no debug entry
in the logfile of pgpool...?!?!?!
why doesn't pgpool recognize that postgres isn't running on host "one" ?
Am I missing something..?!?!
any help appreciated....GERD....
pgsql-admin by date
|Next:||From: Michael Monnerie||Date: 2009-01-23 12:57:07|
|Subject: Re: Why is that index not used?|
|Previous:||From: Chander Ganesan||Date: 2009-01-22 21:52:13|
|Subject: Re: PostgreSQL's query caching behaviour|