Andreas M. Iwanowski a écrit : > Laurent, I don't know if this is a type, but the master host is the one with > the _lowest_ priority.
in keepalived ? extract from `man keepalived.conf` : # for electing MASTER, highest priority wins. # to be MASTER, make 50 more than other machines. priority 100 and in zenloadbalancer: /Priority – connections to the highest priority available/. Balance all connections to the same highest priority server. If this server is down, the connections switch to the next highest server. With this algorithm you can build an Active-Pasive cluster service with several real servers. ???? I m sorry i do'nt understand your remark ? > > -Andy > > -----Original Message----- > From: Laurent Bardi [mailto:[email protected]] > Sent: Monday, April 16, 2012 8:28 AM > To: Emilio Campos; [email protected] > Subject: Re: [Zenloadbalancer-support] cluster with 3 machines ? > > Emilio Campos a écrit : >> Laurent can you be more specific? Where is the sense? > Well, at first, i came from the Keepalived world, so i m not familiar with > Carp/Ucarp. > > I use 3 VMWare ESX servers. Into those servers i use a lot of redundant > virtualized machines/services. > (IE for ldap i ve 3 hosts (virtualized machines) one by ESX server). > > Each host has it own VIPs, one which manage the RIP of the same host(with the > highest priority) > 2 other VIP which manage the others host in case of failure. > > (each host is in fact the service server+load-balancer, > 3 VIP : 10.64.100.2 10.64.100.12 10.64.100.22 for 3 hosts with RIP > 10.64.101.2 10.64.101.12 10.64.101.22 > so for the first host (RIP = 10.64.101.2) , this host manage 3 VIP : > 10.64.100.2 (master, prio = 150) > 10.64.100.12 (backup1, prio = 140) > 10.64.100.22 (backup2, prio = 130) > for the second host (RIP = 10.64.101.12) , this host manage the same 3 VIP : > 10.64.100.2 (backup2, prio = 130) > 10.64.100.12 (master, prio = 150) > 10.64.100.22 (backup1, prio = 140) > for the third host (RIP = 10.64.101.22) , this host manage the same 3 VIP : > 10.64.100.2 (backup1, prio = 140) > 10.64.100.12 (backup2, prio = 130) > 10.64.100.22 (master, prio = 150) > > With this scheme, when one machine fails, one of the 2 other (with the > highest prio) take the hand. > For named services, i've in my DNS the same IN A record (Round-Robin > load-balancing) (IE : > server IN A 10.64.100.2 > server IN A 10.64.100.12 > server IN A 10.64.100.22 > ) > > As you notice, i use RFC 1918 for my IP, and it is compatible with NAT (IE > 10.64.100.2 10.64.100.12 10.64.100.22 are NATed with 3 real IP address of my > network) > > Now, i've found Zenloadbalancer (really nice job!) and i was thinking of > putting 3 special hosts, (one in each ESX) which will make the whole job of > load-balancing/fail-over. > > So, if you can tell me the trick :-} > Many thanks by advance > > (PS: sorry for my English, i m a bloody frenchy :-}) > > > > -- > Laurent BARDI / RSI CNRS-IPBS / CRSSI DR14 INSTITUT de PHARMACOLOGIE et de > BIOLOGIE STRUCTURALE > Tel : 05-61-17-59-05 http://www.ipbs.fr/ > Fax : 05-61-17-59-94 Laurent.BardiATipbs.fr > CNRS-IPBS 205 Route de Narbonne 31400 TOULOUSE FRANCE ... > J'étais indéniablement misanthrope. > Je voulus traverser à gué un marigot infesté d'imbéciles. > Quand j'atteignis l'autre rive, j'étais devenu philanthrope. > > > ------------------------------------------------------------------------------ > For Developers, A Lot Can Happen In A Second. > Boundary is the first to Know...and Tell You. > Monitor Your Applications in Ultra-Fine Resolution. Try it FREE! > http://p.sf.net/sfu/Boundary-d2dvs2 > _______________________________________________ > Zenloadbalancer-support mailing list > [email protected] > https://lists.sourceforge.net/lists/listinfo/zenloadbalancer-support > > ------------------------------------------------------------------------------ > For Developers, A Lot Can Happen In A Second. > Boundary is the first to Know...and Tell You. > Monitor Your Applications in Ultra-Fine Resolution. Try it FREE! > http://p.sf.net/sfu/Boundary-d2dvs2 > _______________________________________________ > Zenloadbalancer-support mailing list > [email protected] > https://lists.sourceforge.net/lists/listinfo/zenloadbalancer-support -- Laurent BARDI / RSI CNRS-IPBS / CRSSI DR14 INSTITUT de PHARMACOLOGIE et de BIOLOGIE STRUCTURALE Tel : 05-61-17-59-05 http://www.ipbs.fr/ Fax : 05-61-17-59-94 Laurent.BardiATipbs.fr CNRS-IPBS 205 Route de Narbonne 31400 TOULOUSE FRANCE ... J'étais indéniablement misanthrope. Je voulus traverser à gué un marigot infesté d'imbéciles. Quand j'atteignis l'autre rive, j'étais devenu philanthrope. ------------------------------------------------------------------------------ For Developers, A Lot Can Happen In A Second. Boundary is the first to Know...and Tell You. Monitor Your Applications in Ultra-Fine Resolution. Try it FREE! http://p.sf.net/sfu/Boundary-d2dvs2 _______________________________________________ Zenloadbalancer-support mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/zenloadbalancer-support
