Thanks to: Larry Clegg and Joe Fletcher.
Within moments of sending my note, Larry confirmed my suspicion with:
-------------
Once you register something with CAA then only root has control of it.
We have many many Oracle databases here under CAA control. If a DBA needs
to do some work on a database then I first do a caa_stop on it. The DBA
does his thing and when finished I do a caa_start. No other way around it.
--------------
Joe had an interesting twist that probably isn't available to us:
-------------------
I've got the same situation at my site albeit on an HP Serviceguard rig. In
our case we had the option to disable monitoring of the database so the
whole package doesn't fail over if we fiddle with Oracle. The end result is
that now we will only get failover if there's a hardware problem.
Not ideal but we live with it.
-------------------
Original note:
We're migrating from 8200s running 4.0f,pk6, ASE, oracle 8.0.6, "manual"
failover of the databases, to ES40s running 5.1,pk3,ase, oracle 8.1.7,
with "automatic" failover of the databases. We currently only have
one database configured in the "automatic" failover mode and the
database and listener can be moved back and forth between nodes and
shutdown using the CAA tools without any problem (if you're the unix
admin). Our DBA wanted to do some maintenance tasks the other day
so she went into Oracle's "svrmgrl" and shutdown the database. The
problem is that ASE detected the database was down and promptly
brought it up on the other system. Are there any easy ways to keep
the DBA and ASE from interfering with each other? From my initial
look at this, it seems we can either give the DBA access to the
CAA tools (less than ideal from the DBA's point of view), or we
have to give up the "automatic" failover. Has anyone else come
up with a solution to this?
Jeff Beck
jbeck_at_carewiseinc.com
206.749.1878
CareWise Inc.
1501 4th Ave.
Suite 800
Seattle, WA 98101-1629
Received on Thu Nov 15 2001 - 20:20:09 NZDT