Network Communication Errors

Performance Manager is a distributed application that depends on NCS RPC (Network Computing System, remote procedure calling). If your RPC environment is not set up correctly, Performance Manager will not work; you will get communications errors for one or more nodes. The following steps should correct a faulty RPC environment on a node:

  1. Verify that the node's host name is correct using the hostname command; for example, on node rintin:
       % hostname
       rintin.zso.dec.com
    

    The output from the preceding command should be the Internet host name of that node.

  2. Determine the host name lookup mechanism being used on that node; there are three: the /etc/hosts file, NIS, and BIND (Berkeley Internet Name Domain).

    To find out which ones are used on a Digital UNIX node, look in the file /etc/svc.conf for the line starting with hosts=; for example:

       % grep hosts= /etc/svc.conf
       hosts=local,bind,yp
    

    Here, local means the /etc/hosts file, bind means BIND, and yp means NIS (formerly yellow pages). The node might use one, two, or all three host name lookup mechanisms.

    The easiest way to determine a node's host name lookup mechanism is to ask your site's network administrator; if that is not possible, you can assume all host name lookup mechanisms are used on all nodes when you perform the following step.

  3. Consult your site's network administrator if you are not sure of a node's IP address, then verify that your host name resolves to the correct IP address using every host name lookup mechanism you found in the previous step:

    For /etc/hosts (local):
       grep  your_host  /etc/hosts
    

    On node rintin, for example:

       % grep rintin /etc/hosts
       16.64.0.99   rintin.zso.dec.com   rintin
    
    For BIND (bind):
       nslookup  your_host
    

    On node rintin, for example:

       % nslookup rintin
       Name: rintin.zso.dec.com
       Address: 16.64.0.99
    
    For NIS (yp):
       ypcat \-d  your_yp_domain  hosts | grep  your_host
    

    your_yp_domain is the domain name returned by the domainname command; see the domainname(1yp) reference page.

    On node rintin, for example:

       % ypcat \-d zso.dec.com hosts | grep rintin
       16.64.0.99   rintin.zso.dec.com   rintin
    

    If any host name lookup mechanism returns the wrong IP address, it must be corrected; consult your site's network administrator.

  4. Verify that your local TCP/IP interface is configured correctly:
       % /etc/ifconfig lo0
       lo0: 127.0.0.1 netmask ff000000 flags=0x9
    

    If your values are different, the TCP/IP interface is configured incorrectly on that node; consult your site's network administrator.

  5. Ensure that the line /sbin/ifconfig lo0 127.0.0.1 appears in these three files:
       /sbin/rc0.d/K50inet
       /sbin/rc2.d/K55inet
       /sbin/rc3.d/S00inet
    

  6. Verify that your network TCP/IP interface is configured correctly. (ln0 is the code for the Ethernet device used in DECstations(TM). For other node types, consult your site's network administrator):
       % /etc/ifconfig ln0
       ln0: 16.64.0.99 netmask ffffff00 flags=0x463
       broadcast: 16.64.0.255
    

    If your output is different (except for the IP address, netmask, and broadcast), the TCP/IP interface is configured incorrectly on that node; consult your site's network administrator.

    In this example, 16.64.0.99 is the node's IP address, which should match the IP address from previous steps. The netmask value and broadcast address depend on the type of subnet addressing used at your site; consult your site's network administrator. The RPC software will not work correctly if either of these is incorrect.


    Back to the User's Guide home page