Topic: Loosing network connection -SOLVED

[drfresh]

I've been using SME 5.6 for years and for a while ago I upgraded my HW and to 7.0.

But I have a problem. When SME 7.0 has been running for some time (less than 24 H - but haven't the exact time) it looses it's network. I can't remotely connect, and when I try to test internet connection in the admin menu directly on the consol there is none.

I'm anything but an expert on this. But is there some hints to some logfiles I can test? I have been using the HW for other porposes (Win XP) for some time and haven't had any ploblems at all with network nor any other topics.

Thanks for any help!

[dexterh]

I tried the upgrade and saw a NIC problem and did a fresh install over it. I'm also losing Eth1 (external NIC).  After reboot, ifconf shows both NICs, but if I do a network restart, it loses eth1.  Using matching RTL8139's on a cheap Compaq P4 Celeron desktop.  

Also having a problem with my 2nd HD.  I can put the second drive in a  S/W raid, however, after next boot, the RAID is no longer there (only one drive shows up -- at least it boots).  Both drives are 60G Maxtor's and  are about a year old.

Wanted to put my RTL8169's back in that I used in the SME 6.5 version.  I had compiled the drivers to get it to work on the 6.5, but if the machine  can't run with a 100 mps NIC, a gigabit one would deffinately be out.  Also see that the SME 7.0 stock RTL8169 drive doesn't work very well according to the posts.

Am running an old Pentium I with SME 5.6 until I can get it to play nice.  If there's a bug, would like to know.  Like the 7.0 features, but the older version's reliability.

[CharlieBrady]

I tried the upgrade and saw a NIC problem and did a fresh install over it. I'm also losing Eth1 (external NIC).  After reboot, ifconf shows both NICs, but if I do a network restart, it loses eth1.

I don't know exactly what "do a network restart" means, but it's not a standard SME server procedure, and behaviour is undefined.

You might get more help if you describe how you are connecting to the Internet. Static IP? DHCP? ADSL PPPoE?

What does "ethtool eth1" say (when you lose connectivity)? "netstat -ieth1"? "ifconfig eth1"?

If you are using PPPoE, then /var/log/pppoe/current is relevant.

[dexterh]

I checked to see if both nics were running upon boot. Initially, they appeared to be working.

When I noticed that only eth0 & lo was showing up with ipconfig after connectivity seemed to stop, I did a "/etc/init.d/network restart" to try to restart the network cards. As I recall it didn't want to do a "service network restart".  Seems Centos4, like Debian 3.1 Sarg has lost the /rc.d to start deamons, though with the SME's, I've not had to do much of anything unless installing rpms or where something needed to be compiled.

At this point, it is still dropping the network connection.  The only way to get limited connectivity is to reboot it, but it's still dropping the external eth1 connection.

I thought it worked after the full "clean" reinstall, however, I did do a "yum update". I'm wondering if I need to do a reinstall without an update to see how it reacts. If so, I'll hold off on any updates until things settle out.

[dexterh]

I noticed another post about seeing the "smp" in the splash screen and did a reboot.  I'm seeing an "smp" on the splash screen.  This looks like the yum update may have jinxed the install.

Will have to reinstall to see if it takes care of the problem.

[drfresh]

Hey man, dexterh - you took over my post! That isn't nice  

Well I still need an answer to my question.

I have an static external IP, static IP on the SME. I didn't do an opgrade of my 5.6 (might have missled you there). I did a fresh install on new HW.

Please come back to my simple question.

[dexterh]

Sorry about hijacking your post.  I thought you were having the issue with the 7.0 version and losing your external NIC.

Just a few things I have seen on other posts and look for are as follows:

1. Is the network running.  I don't recall any good way of doing from the server-manager in 5.6.  I would open a telnet or ssh connection to the server using putty, do a "/etc/rd.d/init.d/networking status" (might want to check the directory to see if it's networking or network - I just type in netw and hit the tab)  I'm pretty sure that the rc.d directory is still used in 5.6.  This status tells you if the network deamon running.  If so, check the NICs.

2. Are the NIC's found and running.  From the telnet or ssh connection to the server using putty, do a "/etc/rd.d/init.d/ifconfig" This tells you if the network cards are running as well as their static IP addresses, dropped packets, errors, etc..  If your having a problem seeing it all, add a " |more" (skip the " marks) after the ifconfig so it limits the screen to a page so you can read the first NIC. This should tell you if they are running.  You would see and eth0, eth1 and lo.

3.  If you look in the /var/log/dmesg log, it may tell you if your hardware is being found and running.

My experience is that the 5.6 version has always worked well for me, however, I've always used matching NICs, as recommended and knock on wood, not had one die.  

Another thing to check is if the motherboard has an on-board NIC and to disable it in the BIOS if you're not using it.  If they all use the same driver, you might not know it's doing anything to you without seeing how many eth's you have. The ifconfig would give you a hint if you see an eth2.

Hope this helps and that someone who may have seen a problem with the 5.6  version can add more info and sorry about stealing your post.  Unfortunately, my bet would be that Charlie Brady was the best bet for figuring it out since he's one of the more kowedgeable folks on SME.  

Might want to also do the "netstat -i" and post as Charlie asked to see if my cards are transmitting and receiving, but it's your cards that may be having an issue.  netstat -i  gives similar information as the ifconfig command, but in table format.

[drfresh]

Sorry about hijacking your post. I thought you were having the issue with the 7.0 version and losing your external NIC.

It's allright!   But it is in the 7.0 I see it - and it's running in server-only mode. And I have no problems usually - but sometimes it just looses the network connection - and in that situation I just can't connect to it. Well I have seen it sometimes, when it has been running for more than 24 H. It's probably an HW problem, though I've never seen it, when it was running WinXP...

Thanks again for any and all help!

[byte]

drfresh,

I removed two of the triple posting you did  

[jongkees]

During the last two weeks I have the same problem. My hardware seems te be correct when I use Knoppix for a test. The server does his job for more than a year.

Sometimes eth0 (server-only mode) is up for 2 full days, sometimes it lives only 2 hours. At that time the server isn't accessible anymore, not via SSH, not even on the console.
After an reboot everything works fine again.

When using ifconfig it sometime shows me no interface at all. Useing 'ifup eth0' bring the interface back up, and I can ping my router again. But all the other network services won't work anymore.

I have no clue if my system is hacked, or there was an update that does this. I tried most of the suggestions above, but without an positive result.  

[ANSW3R]

Same here... I have the same problem...

Is there something to do with the Realtek Ethernet Cards?

[jongkees]

I have an 3COM NIC.
I'am not sure it has anything to do with NIC's. Because sometimes during this situation I do get a ping-responce.

For now the problem (cross my fingers) didn't return. I did a fresh install of SME, and restored all data from the tape-backup. I have absolutely no idea about any update or contrib that could have been the problem, because at the time the craches begon I did not install anything for atleased 3 weeks.

I'll keep you posted.

[drfresh]

Well I still have the problem! The server is running fine, then suddenly I looses connection to the outside world. I can, however, connect to the server via putty, so the netcard isn't dead. When I try to ping from the server and out I see that it can't resolve the addresses I give it e.g. www.contribs.org. It stops receiving mails and so from the outside. I have installed whatever updates that's been avaliable in the server manager.

I usually detect this because I start getting an mail, that "freshclam: update failed"...

Any help would be really appreciated, as this is getting very anoying! Any clues to what to test - I no nothing about linux, so please be specific. Thanks!

[drfresh]

Well it seems that it might have been my gateway (not SME but an router) that was the problem. I've installed a new firmware in it, and I haven't had any problems for some days now. So I'm sorry to have bothered you!

[byte]

Can you put [solved] in subject. Thanks