3 Replies Latest reply on Jan 4, 2013 3:07 AM by alan.pae

    Server can't get to internet

      File server:
      SunFire X4140
      SunOS 5.11 svn_111b


      I'm new to sysadmin, and very new to solaris. Any help would be very much appreciated - I'm trying to get a file server back on the internet that a couple dozen people are waiting for! It's part of a small ad hoc cluster in a rack. It had been working fine every since I started to take care of things part-time 6 months ago. The other day it just dropped off the internet. Then, when I tried a console GUI login at that point it froze up (the GUI, at least). I had to use the serial management port to do a graceful reboot.

      There is an external switch to which all servers are connected, and a local switch used by several (but not all) of the servers in the rack with the file server. I've inherited this system and am unsure of the full rationale of the local switch, but imagine it's for faster local file transfer.

      Other servers in the rack are online, BUT they're running linux variants. The file server runs opensolaris to be able to use zfs. There's another opensolaris server in the rack but it's stuck booting up kubuntu, looking for an NFS mount and I can't get it to skip the mount - haven't focused on this b/c it's non-essential.

      The file server system (see above) boots and I can login, but cannot get online. I get these errors at boot:

      1) The 'opensolaris' graphical boot screen (with the moving orange progress bar) seems to hang. After several minutes I hit return to enter CLI mode. At that point I see errors 2 and 3. I can login.

      2) WARNING: /pci@0,0/pci10de,cb84@2 (ohci0): Connecting device on port 2 failed

      I'm not sure if this is the built-in NIC's, or built-in rear USB ports - or something else. Haven't been able to determine for sure. The rear USB ports don't seem to be working - I have to plug the keyboard into the front usb port to get it to work, so maybe it's the usb.

      3) in.ndpd[429]: interface nge[0|1|2|3] has been removed from the kernel. in.ndpd will no longer use it

      The server has 4 built-in ethernet ports, no others. There are three ethernet cables plugged in - two for the external switch, one for the internal switch. The error for a given interface is only generated when there is a cable plugged into the port, and regardless of whether it goes to local or external switch. Then it repeats the error every minute for 2 of the 3 cables - haven't found a pattern as to which.
      If I connect only one cable, either to internal or external switch, I don't get this error. But, I still don't IPv4 address, see below.

      "ifconfig -a" output (transcribed)
      (cables are plugged into ports 1,2 and 3 (as labeled on the hardware) but not 0):

      lo0: flags=2001000849 <skip>
           inet netmask ff000000

      nge0: flags=1004803<UP,BROADCAST,MULTICAST,DHCP,IPv4> mtu 1500 index 2
           inet netmask ff000000
           ether 0:21:28:3d:6d:2

      nge1: flags=1004843<UP,BROADCAST,RUNNING,MULTICAST,DHCP,IPv4> mtu 1500 index 3
           inet netmask ff000000
           ether 0:21:28:3d:6d:3

      nge2: flags=1004843<UP,BROADCAST,RUNNING,MULTICAST,DHCP,IPv4> mtu 1500 index 4
           inet netmask ff000000
           ether 0:21:28:3d:6d:4

      nge3: flags=1000842<BROADCAST,RUNNING,MULTICAST,IPv4> mtu 1500 index 5
           inet netmask ff000000
           ether 0:21:28:3d:6d:5

      lo0: flags=2002000849 <skip>
           inet6 ::1/128

      nge[1|2] [changes to the nge that was not most recently reported as removed by in.ndpd error.]: flags=2004841<UP,RUNNING,MULTICAST,DHCP,IPv6> mtu 1500 index [3|4]
           inet6 fe80::221:28ff:fe3d:6d0[3|4]/10
           ether 0:21:28:3d:6d:4

      Phew. Thanks very much for any help or suggestions!


      Edited by: 979786 on Jan 3, 2013 11:04 AM

      Edited by: 979786 on Jan 3, 2013 11:08 AM

      Edited by: 979786 on Jan 3, 2013 11:10 AM

      Edited by: 979786 on Jan 3, 2013 12:09 PM
        • 1. Re: Server get to internet
          There might be some eventual replies to your inquiry, but you may need to realize that an online forum is the worst place to hope for help for business-critical issues. The anonymous nature of such a forum is that there is no obligation that there ever be a reply and there is no guarantee that replies are correct for your issue.

          Such a system as you have described needs to be on a support contract (even though OpenSolaris isn't a supportable OS -- it's a discontinued Open Source experiment ). You need to use your support contract to contact Technical Support and open a Service Request to figure out why you have no IPV4 IP addresses, nor any current IPV6 IP addresses.

          ... but I do give you props for at least trying.
          • 2. Re: Server get to internet
            I hear you, thanks for taking the time to reply. We'll see what happens, I'm going for whatever I can. I think my bosses here view me as the service contract...we'll see how well that plays out in the end!

            • 3. Re: Server get to internet
              OpenSolaris should have never been used in a production environment. The license for Solaris itself allows for Oracle to audit you for license compliance so don't try to just brush it off. Get a contract and then have it serviced.