Cluster alias and UDP responses

From: Adrian Liu (adrianliu324_at_hotmail.com)
Date: 06/24/03

  • Next message: WM-data - Tina Stevnhoved: "SUMMARY: numbers of inodes" 
    Date: Tue, 24 Jun 2003 05:37:42 +0000
To: tru64-unix-managers@ornl.gov

    We have an interesting situation with a V5.1 cluster, a cluster alias

    We have a client application that attempts to communicate with a service
    running on a member of the cluster via a UDP port. The service receives a
    client request and sends a response to the client.
    When monitoring the traffic between the client and server using tcpdump we
    see that the initial request from the client comes in and is targetted for
    the cluster alias (which is correct). The response however looks to come
    from the specific member node rather than the cluster alias (which is
    incorrect).

    NOTE: This only occurs for clients that are not in the same subnet as the
    server. For those clients that are in the same subnet as the server, the
    response (reported by tcpdump) comes from the cluster alias, which is
    correct.

    The following tcpdump output shows the situation from client (s-underdale29)
      to cluster alias (mfgprotest), with the member node being pha301. The
    response is coming from the member node which is incorrect:
    16:47:53.724736 s-underdale29.faulding.com.au.1140 > mfgprotest.5162: udp 62
    16:47:53.737431 pha301.5162 > s-underdale29.faulding.com.au.1140: udp 119
    (DF)

    S-underdale29 is 10.25.250.23
    Mfgprotest is 10.3.231.233
    Pha301 is 10.3.231.235

    The following tcpdump output shows the situation from a DHCP client (on the
    same subnet) to cluster alias (mfgprotest). The response is coming from the
    cluster alias which is correct:
    16:55:11.548381 10.3.230.112.4168 > mfgprotest.5162: udp 62
    16:55:11.633340 mfgprotest.5162 > 10.3.230.112.4168: udp 119 (DF)

    Does anyone know how I can test this type of UDP conversation with standard
    Unix? Ie. Remove the application from the picture, incase the application is
    at fault?

    Any other suggestions?

    Thanks,

    Adrian

    _________________________________________________________________
    Find gifts, buy online with MSN Shopping. http://shopping.msn.com.sg/

  • Next message: WM-data - Tina Stevnhoved: "SUMMARY: numbers of inodes"

    Relevant Pages

    • Re: Client/Service relationships & Flow of Requirements.
      ... the client calls "foo" when it wants "bar" to return a particular value. ... The following is designed to compel Prey objects to do something: ... predator < 100 yards from prey ... It is the expectation of the response that makes the runFrom name immediately suspect in an OO context. ...
      (comp.object)
    • Re: [Full-disclosure] [Professional IT Security Providers -Exposed] PlanNetGroup ( F )
      ... the review a second time and incorporate some of your suggestions. ... to do what the client will pay him for. ... exactly a vulnerability assessment is... ... Your response to question 3: ...
      (Full-Disclosure)
    • Re: Debain on the rise ! - However ....
      ... >> Maybe I'll be able to give Debian a try at some point in the future, ... Most threads get response. ... > client shouldn't bitch that they get burned by volume. ... > chance HTML email has content, ...
      (Debian-User)
    • Re: Socket write behaviour is inconsistent?
      ... copy 1 byte to buffer, copy many bytes to buffer, copy one byte to ... Then why did you write "the client throws an error"? ... remote endpoint for your connection. ... When the response is sent using the first chunk of code, ...
      (microsoft.public.dotnet.languages.csharp)
    • Re: Message-based vs. method-based interaction [was: Re: LSP and subtype]
      ... That means the sender implementation does not to know about the response, much less depend in any way on what the responder does, so LSP becomes academic. ... Separation of message and method allows one to design both the client and the service independently. ... Then the entire LSP issue comes down to determining where in the subclassing tree the client should access. ... To do that one needs the OOA/D mindset that clients generate messages, ...
      (comp.object)