IDS 10 performance degrades with dual NICs

Hi we are setting up an Informix server that needs to have dual NICs installed server is fine with 1 card installed but responce times drops dramatically when second card is installed inspite of host being addressed by ip address and not name. Anyone has a clue as to a fix?

What do you mean by "response time"? a) Queries run slower? b) Connect times are slower? c) Both I'd bet that it's b) and if so, it surely has something to do with reverse DNS (which is done by Informix for each connect request) Possibly the machine is not prepared to resolve the IPs that arrive on the second NIC. And it takes time to figure that out. Please note that not being able to reverse DNS is not a problem. Being slow in finding that is the problem. Regards. On Thu, Aug 18, 2011 at 8:47 PM, REECE KONG <reecek@rmp.com.jm> wrote: > Hi we are setting up an Informix server that needs to have dual NICs > installed > server is fine with 1 card installed but responce times drops dramatically > when second card is installed inspite of host being addressed by ip address > and not name. Anyone has a clue as to a fix? > > > > ******************************************************************************* > Forum Note: Use "Reply" to post a response in the discussion forum. > > -- Fernando Nunes Portugal http://informix-technology.blogspot.com My email works... but I don't check it frequently... --20cf3002555ab7a82204aacec306

Since your version 10 you won't be able to add an additional msc vp, but here is an article that may help. Troubleshooting Slow Network Connection Attempts to IDS Users connecting over the network who normally experience split second connection times may suddenly see their connection time become elongated. Applying a tweak here and there can resolve this issue very swiftly the majority of the time (providing an OS or database defect is not the root cause). Before contacting technical support, try applying these few hints. If the problem is noticeable for users authenticated by PAM or for a large amount of users connecting via a connection pool, try allocating an additional ‘msc’ vp. The msc vp is responsible for the execution of operating system calls initiated by other vp’s. These include system calls such as gethostbyname() , gethostbyaddr() etc. This additional allocation is carried out like any other vp, by issuing an onmode –p +1 msc or adding the following to your IDS configuration file “VPCLASS msc,num=2” (adding to the IDS configuration file makes this change permanent). If this fails to resolve the situation, then check with your system administrator regarding recent changes to DNS lookup. Many times, simply changing the lookup order of the remote host , or adding a remote servers i.p. address and name to the database servers local hosts file improve things Also try isolating if the slow connections are only occurring on one service port. If so, try switching to another service port. Finally , profiling the listener thread on the database server may show what system call is responsible for the slowdown: onstat –g ath | grep tcplst and take note of the thread id in the first column. then onstat –g stk -r 1 tread id and look for persistent OS system calls in the callback stack such as gethostbyname(), gethostbyaddr() , getpwuid() etc. If these calls are constantly showing in the callback stack, then the slowdown is in that particular area. Collaborating with your system administrator in helping isolating root cause as well as contacting technical support for additional assistance is then recommended.