On 2/8/07, Daniel Watts <d@xxxxxxxxxxxxx> wrote: > Paul Lesniewski wrote: > > On 2/2/07, Daniel Watts <d@xxxxxxxxxxxxx> wrote: > >> > >> > >> Alexandros Vellis wrote: > >> > >> Thank you for this - we've just found out that haproxy can set a cookie > >> on the client side that tells the load balancer which server to connect > >> to. HAproxy is clever enough to know that if it tries to connect to the > >> same server and fails, it ignore the cookie and connect to another server. > >> > >> A hah. Sounds nice. > >> > >> "pound" has a similar feature, and it can actually maintain the session > >> mapping itself without any additional cookie on the client. > >> > >> Yep, so does HAproxy but what happens if the client connects to the OTHER > >> load balancer during the same session? Either the client needs to have some > >> state information (a cookie) or the two servers need to share some state > >> information. haproxy doesn't have the ability to send such information > >> between instances so we had to go for client side state tracking. Does Pound > >> have this facility? > > > > Last I set up a system like this (using Ultra Monkey), we just set > > connection persistence and away we go. This has already been > > suggested, I know, but I will 2nd the suggestion. It works like a > > charm. > > Hi Paul, > > Thanks for the confirmation. Again though - if you have two Ultra Monkey > servers under roundrobin DNS, how do you ensure that whether UM server > the client hits, their session is maintained? Can UM servers talk to > each other? I don't know, but it's a good question. I'm wondering if you wouldn't use some kind of failover system instead (if you are trying to have redundancy for the load balancing server), since it seems odd to me to have a RR system feeding what might be another RR system. Then doesn't your DNS server become a single point of failure (although I suppose the fact that you can register more than one answers that question, but read DJB's opinion on the fallacy of needing multiple DNS servers in different locations)? Eventually you might have to eat the fact that something can break and your systems will go down. This might be best asked in another place more specific to these kinds of systems. I'd be very interested in the answer if you find one. > I've been looking at UM actually - I like it's custom 'negotiation > checks' that can do more than simply pinging a port or checking a webpage. ------------------------------------------------------------------------- Using Tomcat but need to do more? Need to support web services, security? Get stuff done quickly with pre-integrated technology to make your job easier. Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642 -- squirrelmail-users mailing list Posting Guidelines: http://www.squirrelmail.org/wiki/MailingListPostingGuidelines List Address: squirrelmail-users@xxxxxxxxxxxxxxxxxxxxx List Archives: http://news.gmane.org/thread.php?group=gmane.mail.squirrelmail.user List Archives: http://sourceforge.net/mailarchive/forum.php?forum_id=2995 List Info: https://lists.sourceforge.net/lists/listinfo/squirrelmail-users
