Re: monitor rac database services
- From: Yong Huang <yong321@xxxxxxxxx>
- To: Freek.DHooge@xxxxxxxxx, coskan@xxxxxxxxx
- Date: Wed, 15 Jun 2011 14:13:19 -0700 (PDT)
> I know the script, but I don't think it takes the target state of
> the service in account. Which means an alert would be given if a
> service is down, even if the target state for that service on that
> node is "offline".
That's correct. I wrote the script
(http://yong321.freeshell.org/oranotes/Service.txt) originally for
10g and it works for 11g too, but it doesn't consider TARGET, i.e. what
clusterware thinks it should be online or offline, because my DBA
group is only interested in STATE, as well as whether the service is
running on its preferred instances.
As far as I know, in 10g, only crs_stat shows TARGET. In 11g, both that
and "crsctl stat res" have that. But in 11g RAC, crs_stat is no longer
reliable about service statuses (for example, after you stop a service
with dbms_service.stop_service, or crash the instance, crs_stat still
says it's ONLINE). The preferred command "crsctl stat res -t" (or
without -t) is reliable. I'm not aware of any way to get TARGET from
inside database. Although x$kswsastab, base table of v$(active_)services,
has a few columns not exposed, I doubt it has TARGET; they may be some
less commonly used attributes as shown in "srvctl config service" in 11g.
But I could be wrong.
I thought all this info was in OCR. But ocrdump, even run as root to dump
everything, doesn't have it. Also checked the output of "ocrconfig -export".
Yong Huang
--
//www.freelists.org/webpage/oracle-l
Other related posts:
