Federico, 503 would be due to a node being not responsive, you can configure your cluster to disable a node after one or more failures with mod_cluster. now back to mod_proxy_balancer and mod_proxy, if i understand correctly, you have a scenario where you have one or more contexts served by all of the nodes in the cluster, and another context served only by some nodes, i.e.:
Node1 => /foo,/bar Node2 => /foo Node3 => /foo,/bar you want /foo to be served by all of the nodes, and /bar just by Node1, and have Node3 as a hot standby (failover), if that is the case, you would probably need multiple load balancers defined with two different location blocks, so something like this: <proxy balancer://foocluster> BalancerMember http://node1:8080 BalancerMember http://node2:8080 BalancerMember http://node3:8080 </proxy> <proxy balancer://barcluster> BalancerMember http://node1:8080 # the hot standby on node3 BalancerMember http://node3:8080 status=+H </proxy> <location /foo> ProxyPass balancer://foocluster ProxyPassReverse http://node1:8080 ProxyPassReverse http://node2:8080 ProxyPassReverse http://node3:8080 </location> <location /bar> ProxyPass balancer://barcluster ProxyPassReverse http://node1:8080 ProxyPassReverse http://node3:8080 </location> i haven’t tested above configuration. -Roman On Jul 3, 2014, at 10:48 AM, Jeff Trawick <traw...@gmail.com> wrote: > On Thu, Jul 3, 2014 at 9:57 AM, Federico Calì <f.c...@tdnet.it> wrote: > Hi Roman, > the mod_cluster is what we have actually in production. > I want to change it, because the communication > between Jboss and front-end sometimes fails with error 503. > I would like to use ProxyPass in order to investigate if these kinds of > errors will remain. > > It is very likely that you'll have the same problem, which is an issue, > perhaps intermittent, with the backend server or application. > > Maybe one proxy implementation reacts a little better in your circumstance > and/or can be configured to react better (e.g., recover more quickly), but > you'll probably end up needing to diagnose the root cause either way. > > > Federico > > > > Il 03/07/2014 15:15, Roman Jurkov ha scritto: >> Federico, >> >> since you are using JBoss, i would suggest to look into mod_cluster instead. >> >> http://mod-cluster.jboss.org >> >> -Roman. >> >> On Jul 3, 2014, at 8:58 AM, Federico Calì <f.c...@tdnet.it> wrote: >> >>> Hi Eric, >>> >>> thanks for the hint but I think it's not what I need. >>> Defining the standby node(s) in the balancer definition I might loss >>> functionality for these reasons: >>> >>> - the JBoss cluster has both web applications deployed in HA Singleton that >>> deployed in "statndard classic" [ie running on all nodes]; >>> For the second kinds of applications each request can be served by all >>> back-end node. >>> >>> - I can't know which node is the HA Singleton Master; >>> >>> - I might need to extend the jboss cluster adding new nodes (2,3,..) but >>> only one will be the master for HA Singleton web applications; >>> >>> Federico >>> >>> >>> Il 03/07/2014 13:59, Eric Covener ha scritto: >>>> On Thu, Jul 3, 2014 at 6:47 AM, Federico Calì <f.c...@tdnet.it> wrote: >>>>> I'd like use Apache ProxyPass, with balancer directives, to refer these >>>>> web >>>>> applications, >>>>> but I need that no requests are sent to the "stand-by" back-end node. >>>> http://httpd.apache.org/docs/2.2/mod/mod_proxy.html >>>> >>>> Search for "standby" >>>> >>> >>> -- >>> >>> .: Federico Calì >>> .: TD Group S.p.A. >>> .: Via del Fischione 19 >>> .: [Via Traversagna -Ingresso Zona C ] >>> .: 56010 Migliarino Pisano (PI) >>> .: 050-897406 >>> .: 348-2886628 >> >> Nessun virus nel messaggio. >> Controllato da AVG - www.avg.com >> Versione: 2014.0.4592 / Database dei virus: 3986/7788 - Data di rilascio: >> 03/07/2014 >> > > -- > > .: Federico Calì > .: TD Group S.p.A. > .: Via del Fischione 19 > .: [Via Traversagna -Ingresso Zona C ] > .: 56010 Migliarino Pisano (PI) > .: 050-897406 > .: 348-2886628 > > > > -- > Born in Roswell... married an alien... > http://emptyhammock.com/ > http://edjective.org/