Hello, We spotted the cause of instability: The pool machines had been previously tuned at their tcp kernel parameters for high speed networks. One of the tcp parameters tuned caused the problem with pgpool and postgresql under heavy loads (backends at 100%): tcp_fin_timeout was at only 10 seconds. After configuring it to the default linux 60 seconds, the connection instability under heavy load disapeared. At least for some days now. Regards. Andre Felipe Machado
_______________________________________________ Pgpool-general mailing list [email protected] http://pgfoundry.org/mailman/listinfo/pgpool-general
