Re: Cheetah2 connection manager and failover arbitrator

informix-list-bounces@iiug.org wrote on 06/09/2008 09:54:41 AM: > ----- Original Message ----- > From: "Andrew Ford" <aford@networkip.net> > To: <informix-list@iiug.org> > Sent: Friday, June 06, 2008 12:53 PM > Subject: Cheetah2 connection manager and failover arbitrator > > > > IDS 11.50.UC1 > > Red Hat Linux 5 > > Simple HDR environment 1 Primary - 1 Secondary > > > > Some additional CM questions not related to RAC... > > When the CM is started in a simple HDR environment (or complex > MACH11 env I guess), should INFORMIXSERVER for the environment > starting the CM always be set to the current HDR Primary? If so, No, you can set INFORMIXSERVER either to a primary or any other secondary server in the cluster. > scripting the automatic start of a CM in an environment where the > HDR Primary can change often will need to identify which server is > primary and properly set INFORMIXSERVER before starting a CM, > right? > > If you have a dedicated network for HDR traffic, can you prevent the > CM from using the DBSERVERALIAS for this network for connection > redirection? I think so - depends on how you configure SLA. To avoid redirection to dedicated port, use explicit server names in SLA configuration instead of server type and avoid using the dbserveralias which is configured with s=6. > Setting the s=6 option for both HDR sqlhosts entries > prevents connection attempts on the HDR network for the primary > SLA, which is good. But setting the s=6 option for both HDR > sqlhosts entries prevents ALL redirection to an HDR (secondary only) > SLA with a > > Warning: no server returned for SLA cheetah_sec2 [cmsm_sla.c:1117] > SLA cheetah_sec2 cannot find database server for SQLI client > [cmsm_main.c:1151] > > in the oncmsm log > > And finally > > When running two CMs to provide redundancy, could there be any > issues with both failover arbitrators realizing the primary node is > down and attempting a failover to the secondary at the same time. No - only one CM will act as failover arbitrator. In CM log, check for the message "Arbitrator is active on CM = <cm-name>" Second CM will act as arbitrator only when first CM is down. HTH, Nilesh. > My tests don't show a problem so far, wondering if anyone can > comment on if this condition is handled in the CM. I suppose a way > to avoid the problem is to set the failover timeout value to N > seconds on CM1 and N+2 seconds on CM2. > > Thanks, > > Andrew > > > _______________________________________________ > Informix-list mailing list > Informix-list@iiug.org > http://www.iiug.org/mailman/listinfo/informix-list