I can say for sure that within our environment here, which is Websphere 5.0.1 on Linux, a cluster of about 20 machines (and a number of clusters, but the clusters themselves are independant so it's not relevant here), each request that comes in can be serviced by any of the 20 machines in the cluster. Regardless of which one a request is serviced by, the session is available. Is it writing to a common database? Or maybe has some RMI process working between them to do the session replication? I don't know that answer, although something like the second seems more likely to me, for the very performance reasons you rightly point our Leon. :)
Mapping a drive will work, but the mapping has to be on the app server, not the web server, right? However, this is also something that the admins here do not want to do (because of the whole "you can't write to the file system PERIOD!" argument). But that's not a technical issue so much as a political one :) However, there ARE good arguments for writing to the file system, and the main one is that in a clustered environment, assuming you don't have a mapped drive, you won't know where the file was written to (not without building some logic into your app to identify what server the upload came in on, which is ugly). -- Frank W. Zammetti Founder and Chief Software Architect Omnytex Technologies http://www.omnytex.com On Mon, March 7, 2005 3:33 pm, Dakota Jack said: > Actually, Leon, I think it is really relevant and important to struts > uploading. The point I was trying to make, however, and I think you > will agree, is that there has to be somewhere you put the files if you > allow uploading and there has to be some output stream doing it. That > means you can monitor no matter what you do, right? > > Jack > > > On Mon, 7 Mar 2005 21:22:52 +0100, Leon Rosenberg > <[EMAIL PROTECTED]> wrote: >> That depends. Ok, if you distribute the session, well, then you probably >> don't need a clustered environment, >> but a doctor :-) I mean, one of the goals of the clustering is >> performance, >> and write something in the >> database on _each_ request is the best performance killer :-) >> >> If performance is not an issue, you can mount same filesystem on each >> webserver, and check the progress on >> the filesystem. >> >> At my current project, we are hosting on 12 machines with session >> persistence. This means, that if a machine >> crashes all users logged in onto this server has to relogin. So far it >> happened 5 times (I curse tomcat) in last 6 month. I think it's ok for >> most >> users / websites. Most of our users didn't noticed it or blamed the >> DSL-provider >> or the browser (which is always a good point for flames, by default:-)) >> >> But I think this is getting far far far off topic... >> >> Regards >> Leon >> >> > -----Ursprüngliche Nachricht----- >> > Von: Frank W. Zammetti [mailto:[EMAIL PROTECTED] >> > Gesendet: Montag, 7. März 2005 21:14 >> > An: Struts Users Mailing List >> > Betreff: Re: AW: DownloadAction Application >> > >> > Is that the way it works in your environment Leon? Its not >> > the way it works here :) Or any other clustered environment >> > I've ever worked in. >> > >> > I mean, the express purpose of a clustered environment is to >> > distribute load. If a user hits one server for one request, >> > but then upon making the second request that server they hit >> > the first time is loaded down, the whole point of the cluster >> > (ignoring failover and such of course) is to get that user to >> > a less loaded server so that all users have a good response >> > time on average. >> > >> > That's also the reason IBM tells you to keep session small: >> > it has to be replicated across the cluster so that any >> > request can be serviced by any machine in the cluster at any >> > given time. >> > >> > That's not to say you COULDN'T configure things to always >> > direct a user to a given server. But I think that defeats >> > the purpose of such an environment :) >> > >> > -- >> > Frank W. Zammetti >> > Founder and Chief Software Architect >> > Omnytex Technologies >> > http://www.omnytex.com >> > >> > On Mon, March 7, 2005 3:05 pm, Leon Rosenberg said: >> > >> Assuming so... in a clustered environment, first of all, >> > writing to >> > >> the file system is generally discouraged practice >> > (although it can be >> > >> done safely, so let's ignore what might be best practice for the >> > >> moment)... but if you do so, since the upload may start on >> > one server >> > >> and then the monitor gets executed on any server in the >> > cluster with >> > >> each subsequent check operation, how does that get handled? >> > > >> > > The loadbalancer should send all requests from one user to same >> > > server, so it's not an issue. >> > > >> > > >> > > >> > > >> > --------------------------------------------------------------------- >> > > To unsubscribe, e-mail: [EMAIL PROTECTED] >> > > For additional commands, e-mail: [EMAIL PROTECTED] >> > > >> > > >> > >> > >> > --------------------------------------------------------------------- >> > To unsubscribe, e-mail: [EMAIL PROTECTED] >> > For additional commands, e-mail: [EMAIL PROTECTED] >> > >> > >> >> --------------------------------------------------------------------- >> To unsubscribe, e-mail: [EMAIL PROTECTED] >> For additional commands, e-mail: [EMAIL PROTECTED] >> >> > > > -- > "You can lead a horse to water but you cannot make it float on its back." > ~Dakota Jack~ > > --------------------------------------------------------------------- > To unsubscribe, e-mail: [EMAIL PROTECTED] > For additional commands, e-mail: [EMAIL PROTECTED] > > --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
