eeE: Lost Link syslog error message

[pam196]

Hi All

I'm hoping I already know the answer to this one - it confirmation I'm after really! A mission critical remote server dropped off the network at a given time, and reported "eeE: Lost Link" in syslog. My thoughts are that for the server to have written to syslog, suggests its not a hanging problem / shared memory issue etc. I'm presuming this message is what you would see had the network cable been removed from the back of the server, or patch point in the hub. Has anyone come across this before, are my presumptions correct ?

Phil

 

[DSMARWAY]

Yes you are right
The probable causes of this error could be :-

1. Bad cable
2. Bad Port on hub
3. Bad Network card
4. Bad PCI Slot on the machine the network card plugs in
Have you tried swapping , cable , connect to another port on the hub ,configure another network card if possible
, use another PCI slot etc .?


5. If all above is o.k , have you tried to install the latest driver

 

[pam196]

On a reboot of the server during the problem time, it loaded up with portmapper error messages. On a second reboot, the patch cable into the back of the machine was disconnected and reattached (although, cannot confirm it was attached in the first place - telephone support!). It came back online successfully.

It turns out today, the company providing network support are picking up full duplex errors on the switch / hub - though not for this particular server. Though, if the hub is picking up duplex errors - it could be a possible cause of the loss of link? I can only presume the hub was jabbering to the server and knocked it offline - the fiddling with the patch cable disconnected comms to the hub and cleared the jabber? The second reboot cleared down comms from the server - and hence all OK when back online.

I am somewhat weary of installing the latest driver (this is SCO afterall!); mostly on the grounds that I administer 50+ identical servers (not at the same site) all running the same version driver - all OK.

I think I'll see if the problem reoccurs, and make descisions on hardware from there, after addressing the duplex problem - I still think the most likely cause is that it was accidentally disconnected - as all patch cables are brand new.

 

[DSMARWAY]

Is your hub just a bog standard hub or an intelligent hub /switch? i.e. has it got the capability to set the speed on individual ports ? if so it may be worh checking , that the port on the hub is set to the same seting as the card has, i.e. 10MB full duplex and not 10MB full and 10 Mb half or something similar.

 

[trikk]

Here's an off the wall (yet common) question...in this servers /etc/hosts file...do you have a duplicate ip address entered? teh port mapper error is most commonly (from my expierence) caused by a dup IP enetered here.

If not, I am willing to bet its a bad cable/hub/switch, or Dead nic. Outside shot is the pain in the butt issue of the Intel Pro chipset on some NIC's needs a patch to play nice with teh OS. Have you received any "K-Trap" errors or unexpalinable system reboots recently?"

 

[pam196]

DSMARWAY - I don't manage or have anything to do with the hub unfortunately. I'm getting to the point where I'd rather the hub config changed than some of the servers. E.g. on some NIC's you can't change duplex settings from netconfig. The override method is a little intense - changing c source and header files for the NIC driver.

trikk - there's no duplicate IP's in /etc/hosts; however, you mention a patch for the NIC - I have had a couple of inexplainable system reboots. By inexplainable I mean akin to a powercut with no UPS. No error logging whatsoever my syslog - system as if turned off from mains. Where and what are K trap errors? Presuamabley they'd be logged to syslog? - in which case I haven't identified any. What does the patch identify / fix - and where can I get it from?