>
>
>FYI,
>
>For any other company that is looking to implement the FOUNDRY ServerIron XL
>
>>There is a resolution to this problem...it was not a misconfigured set of
>>ports.
>
>There was a bug in the Foundry code that was fragmenting UDP packets for
>LDAP(Active Directory)....the NT boxes are now behaving normally.
>
> >>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<
>April Sims MCSE, CNE Abilene Christian University
>OCP 8i 9i DBA ACU Box 29005
>Information Technology Abilene, TX 79699
>[Email Address Removed] Vx:(325)674-2681 Fx:674-6724
> >>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<
>
>
>
>
>>After quite a bit of testing we are almost ready to implement load
>>balancing for INB and Web for Products using 2 identical NT servers using
>>9iAS 1.0.2.2.2a
>>We are using the default settings that come with the switch alllowing
>>traffic on ports 80, 443 and 8010.
>>
>>There are a couple of observations that hopefully someone out there has
>>also seen: (Wayne State are you out there?)
>>
>>1. NT Servers while attached to Foundry Load Balancer every new
>>Windows2000 logon takes forever....but once it happens the first time
>>subsequent logins are no problem. Talking about a very long time, take
>>the boxes off the switch then the logons are back to usual.
>>
>>2. When using Oracle WebHammer we notice there is no round robin
>>happening, each connection (out of 50000 hits) stay with the same NT server.
>>If I start up a different WebHammer session with a different client it
>>hits persistently with the second NT server.
>>Now if we do the same hits against dual UNIX boxes with 9iAS it flips
>>back and forth no problem as expected with a load balancing switch.
>>
>>The httpd.conf are the same in KeepAlive, MaxKeepAliveRequests, etc.
>>
>>The second item is the one that really concerns us.
>>
>>Any insight or changes we need to make to the Foundry Switch would be
>>appreciated.
>>Thanks.

Oracle LazyDBA home page