I recently installed Solaris 10 1/13 and proprietary TCP PEP software on a new hard drive within an old Sun v210. When the PEP software starts, the Sun's CE0 interface loses its "ether" address, according to "ifconfig -a".

Here's the output before the software starts:

ce0: flags=1000842<BROADCAST,RUNNING,MULTICAST,IPv4> mtu 1500 index 4
     inet netmask 0 
     ether 0:14:4f:33:82:a1 

Here's the output after the software starts:

ce0: flags=1000842<BROADCAST,RUNNING,MULTICAST,IPv4> mtu 1500 index 4
     inet netmask 0 

The IP address, netmask, and flags are all displaying the expected values. It's just the vanishing MAC address which is a problem. The interface will not pass any data, despite the link being "UP" according to the switch. The server is also using two BGE interfaces which experience no problems whatsoever.

I couldn't find other instances of this issue online, but it could be that it's difficult to phrase the question in a sufficiently specific way to get useful search results. Has anyone else seen this problem in the past? We've used this PEP software on hundreds of v210s over the better part of a decade, and I can find no suggestion that his issue has occurred in the past.

The only thing different about this instance, is the fact that I'm using Solaris 10 1/13; the v210s usually use older releases of Solaris 10, but 1/13 is the only one available from Oracle anymore. Other than rolling back the CE interface driver to match a known-good version, is there anything else related to this interface which may need to be rolled back which could cause these symptoms?

I know this question is probably made more difficult by the inclusion of our proprietary software into the mix, but I appreciate any advice anyone can offer.


There are two directions to choose from:

  • either you go back to a known-good setup (I guess you'll have to find a way to get an older version of Solaris for this, though),

  • or you find out what is the cause of the problem.

My first suspect would of course be the proprietary application, doing something really nasty to that ce interface. You could use truss to trace the application and see the activities.

Last time I've seen these letters (PEP) for an application, it referred to some transparent proxy in the network of a mobile operator, capturing IP traffic of mobile phones and transparently compressing it (changing HTTP headers as necessary). If your application is doing the same, the chances are that the application is messing with the MAC address to be able to do the transparent proxying.

  • Thanks, I'll look into the truss command and see what I can turn up that way. Just for clarification regarding PEP: it's a Performance Enhancing Proxy for a satellite communications system; it spoofs part of the TCP syn/ack cycle, so the server doesn't have to wait through 500ms+ satellite latency to get an ack for each packet. – Liesmith Dec 18 '13 at 0:06

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.