No particular errors in the VCS log, but using VCS online and offline commands like hagrp -online, hares -online, will be in hung state and timeout.
The Symantec High Availability Agent for PostgreSQL does not specify shutdown mode when stopping PostgreSQL server. The shutdown mode may be smart, fast, or immediate and the default mode is smart.
# $sBaseDir/pg_ctl stop -w -D $sDataDir
The shutdown command waits for all the connections to terminate and this causes VCS to call the clean script. Processes could be left behind and not killed.
Workaround is to either terminate those waiting/idle connection processes and/or use PostgreSQL shutdown option, "-m fast". The "fast" shutdown mode does not wait for clients to disconnect. All active transactions are rolled back and clients are forcibly disconnected, then the server is shutdown.
Veritas has released hotfix "vcs_agents-linux-5.1.4.1HF1" with the workaround of "-m fast" shutdown option. Please contact Veritas Technical Support to obtain this hotfix.
Further the next release of PostgreSQL agent (1Q2014 agentpack) will be enhanced to provide a way to specify stop options when stopping PostgreSQL server.
Applies To
RHEL6.1 / VCS 6.0.1 / Postgres 9.2 / VRTSpgsql-5.1.3.0 / VRTSpgsql-5.1.4.0
postgres 311 3.6 0.1 233904 40648 ? Ss 09:00 4:34 postgres: amsclient amsgeo_farm 10.0.0.10(49756) UPDATE waiting
postgres 418 0.0 0.0 244848 23960 ? Ss Jun26 1:04 postgres: amsclient amsgeo_geodatabase 10.0.10.10(39908) idle