postgres-xc-developers Mailing List for Postgres-XC

Hi,

I made the same test as you and didn't notice any errors when dropping the
second database.

By Having a look at the logs of your nodes when you stopped them, it looks
that it entered in recovery mode.
So I assume you stopped it in "fast" mode with pg_ctl or with SIGQUIT.
In this case database shuts down without waiting for the client to
disconnect.

My cluster structure is 2 Coordinators and 2 datanodes, with 2 proxies.
After doing the same SQL, I stopped my nodes in the most violent way.
GTM and proxies were not shut down.

pgsql]$ ./bin/pg_ctl stop -m fast -D ~/pgsql/coord1/
waiting for server to shut down.... done
server stopped
pgsql]$ ./bin/pg_ctl stop -m fast -D ~/pgsql/coord2/
waiting for server to shut down.... done
server stopped
pgsql]$ ./bin/pg_ctl stop -m fast -D ~/pgsql/datanode2/
waiting for server to shut down.... done
server stopped
pgsql]$ ./bin/pg_ctl stop -m fast -D ~/pgsql/datanode1/
waiting for server to shut down.... done
server stopped

Then I restarted the nodes, and dropped a database. It worked.

I did also a second test.
I stopped my nodes, then the proxies (GTM not touched):
pgsql]$ ./bin/gtm_ctl stop -m fast -S gtm_proxy -D
/home/michael/pgsql/proxy2
waiting for server to shut down.... done
server stopped
pgsql]$ ./bin/gtm_ctl stop -m fast -S gtm_proxy -D
/home/michael/pgsql/proxy1
waiting for server to shut down... done
server stopped
I restart the whole cluster, then drop again a dabatase.
It worked once again.

I made a couple of times the same tests and it looked to work all the time.
The logs on each node also looked correct.

Regards,

-- 
Michael Paquier
http://michaelpq.users.sourceforge.net