Hi, All I think that there is a problem for the monitoring of the current pgsql resource. It is two problems.
1)A monitor cannot judge that plural PostgreSQL instance works in the same node. This is because monitor judge it by an instance name of PostgreSQL with pgrep. For example, this problem occurs when consist of plural nodes. When all PostgreSQL is started in one last node, it occurs. 2)pgrep of the monitor cannot judge PostgreSQL instance done swap out. This may not occur very much. But, it cannot be said that it never occurs. I watched the manual. /proc/pid/cmdline seems to disappear in swap out. But, pgrep seems to watch cmdline. For the solution of the problem, a change of pgsql is necessary. For example, 1) Use the PID file as before. 2) But use pgrep when monitor can not see a PID file. 3) Furthermore, I appoint the required parameter that pgsql can judge in pgrep. -> By a parameter, the monitor can judge the plural instance. The situation of the worst swap out may not evade even this example. But, I think that this example is improved than present pgsql. Please show me an opinion of all of you. Best Regards, Hideo Yamauchi. _______________________________________________ Linux-HA mailing list [email protected] http://lists.linux-ha.org/mailman/listinfo/linux-ha See also: http://linux-ha.org/ReportingProblems
