Hi Christopher I don't have access to code inside application, I'm using stock Sonatype Nexus OSS who use Shiro internally to manage (and regenetate JSESSIONID cookie), loosing jvmRoute contents.
See https://issues.sonatype.org/browse/NEXUS-5756 We could told mod_jk to use another cookie but I think Tomcat allways append jvmRoute content to cookie JSESSIONID. That's why Rainer and I discussed about mod_jk to generate and inject another cookie and use it to maintain sticky affinity. It will be an even better solution for large cluster of Tomcat behind HTTPd/mod_jk, especially when instance are automatically provisionned from template since we don't have to set jvmRoute in server.xml for each instance. But I may be wrong and may be Tomcat 7 could inject jvmRoute in another cookie ? Cheers 2013/6/25 Christopher Schultz <ch...@christopherschultz.net> > Henri, > > On 6/25/13 2:33 AM, Henri Gomez wrote: > >> Henri, > >> This sounds more like a question for the users list. Cross-posting to > >> re-locate. See more below... > > > > > > I used dev list because question was also about updating mod_jk code. > > > >> Is is important to retain the original JSESSIONID, or do you only want > >> to retain the jvmRoute information? I would think you'd want both. > > > > jvmRoute should be kept in JSESSIONID as mod_jk load-balancing use it to > > dispatch requests to proper Tomcat :) > > Of course! But it sounded like both Tomcat and Shiro generate a > JSESSIONID but you were only worried about Tomcat's /jvmRoute/ (and not > the rest of the session). I'm assuming that you want to maintain both > session ids simultaneously. > > Assuming you want to retain both at the client, why not simply change > the name of the cookie for Tomcat (or Shiro)? > > -chris > >
