From: Koichi S. <koi...@gm...> - 2013-05-13 07:20:28
|
Hi, I tested the patch and found it works fine. Now I can stop the master first to remove and then issue DROP NODE both for coordinator and datanode. It is much safer. Thank you. ---------- Koichi Suzuki 2013/5/13 Koichi Suzuki <koi...@gm...> > Hello; > > I'm testing pgxc_ctl to remove a coordinator master. Following your > suggestions, I first stopped the coordinator master to be removed and then > issued "drop node" in all the other coordinators. Somehow drop node > complains that it fails to obtain pooled connection, in fact, in drop node, > "execute direct on (node_to_be_removed)" was issued. > > It worked fine if the removed coordinator is stopped after all the DROP > NODE are successful. > > Did you issue CLEAN CONNECTION or pgxc_pool_reload before you issue "DROP > NODE"? > > Regards; > ---------- > Koichi Suzuki > |