Network problem - ubuntu

Hey

Im having network trouble with my ubuntu server 3.19.0-58-generic #64~14.04.1-Ubuntu
The server can be running smooth for one week, maybe one month, but then out of nowhere something is somehow conflicting.

The lan still seems to be working but no devices in the house got any connection at all.
Then If I unplug the server networkcable, reboot the router and reboot the server and replug it, everything is up and running again.

I have set the the ip static for the server, choosen an ip out of the ip poolrange and running dhcp on rest of the devices.
If you need more info please let me know what to find.

Does anyone have an idea how to fix this annoying problem?



provo@plexen:~$ cat /etc/network/interfaces
# This file describes the network interfaces available on your system
# and how to activate them. For more information, see interfaces(5).

# The loopback network interface
auto lo
iface lo inet loopback

# The primary network interface
auto eth0
iface eth0 inet static
address 192.168.1.150
netmask 255.255.255.0
gateway 192.168.1.1
dns-nameservers 8.8.8.8
provo@plexen:~$


Thanks

does the router also act as the switch for you network? or are you using something else.

are there any errors showing in the logs?

Yes the router is also used as switch
I cant find anything in syslog.

I tried using another router, same problem.

Fotex



Edited 1 time(s). Last edit at 04/26/2016 01:46AM by Fotex.

Fotex Wrote:
-------------------------------------------------------
> Yes the router is also used as switch
> I cant find anything in syslog.
>
> I tried using another router, same problem.
>
> Fotex

What type of hardware this Ubuntu server is running on? Do you keep dmesg log on disk?

It's hard to tell what happened exatly at the moment it stops responding without a serial console (I asume you don't have it connected). So next time this occurs, if you can get to the dmesg and syslog in /var/log/, it might show the error.

Or it coulb be something that cause the CPU to go 100% so LAN seems to be alive, but nobody can connect.

Can you still ping the server when this occur?

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)

The server is some how killing the router / entire network, i cant access anything when this happens.
The laptop shows connected to the router but nothing is working. (No ping respond)


The server is an i5-2400

if its on a laptop, my guess is you have a faulty NIC. or some control signal capacitors on the way out/blown.

Try with a seperate usb connected network card or a docking bay if you have one. or if its connected to a docking bay

It could be caused by some programs with some sort of memory leaks that start to drain the available RAM. When a system runs out of RAM, a lot of things can happen.

The server is a headless box placed in a closet.
My laptop is just one of the devices that also looses connection when this happens.

I found the syslog from last time I had the problem and I know it was arround Apr 22 18.~
I have the compressed dmesg from that week, but they got no timestamp. Is there a way to work arround that and locate the wanted file?

first thing i would do i look to resolve the error and warning messages in the log file you posted

+ use an image that does not have a GUI loaded that consumes resources. if you are running this headless there is no need for a GUI

It is ubuntu server, but I have installed gui because Teamviewer needed it run.

> if its on a laptop, my guess is you have a faulty
> NIC. or some control signal capacitors on the way
> out/blown.
>

Given the ping does not work, I tend to agree with Gravelrash that it points to the faulty NIC. Or something that took down the network interface.

However, take the compressed dmesg and uncompress it. Check if there is any error like a failing device. And find out how what type of your NIC hardware is, e.g. is that a PCI card, when the server running OK.

If you suspect that is software related, then you can run top in batch mode, and route the log file to /var/log. If logrotate is running, make sure that it also rotate and compress this top output file. This log should tell you if any process took 100% CPU or 100% memory when this occurred. Example:

top -n 5 -b > /var/log/top_5_seconds.log

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)

Can that cause all other devices to loose internet connection as well?
I tried to grep through the logfiles searching for errors, and found some.
however I dont know if this is something alerting.

Fotex Wrote:
-------------------------------------------------------
> Can that cause all other devices to loose internet
> connection as well?

yes as it could bark allsorts of noise down the line and flood the network which could overload the router you are on

Ok I think ill try and use a pci-card insted of the onboard one, if the logfiles dont show anything at next breakdown.

Thank You All for your response

- fotex

From the logs, it seems you are using a motherboard with Intel e1000e ethernet chip. This chip is currently known for locking up for no reason with kernels 4.X, maybe you got hit by the same bug even with an older kernel (when it was rarer).

http://www.phoronix.com/scan.php?page=news_item&px=E1000E-Crazy-Hangs

Using another non-Intel chip will probably solve the current issue.

Being Intel, this is probably going to get fixed within a reasonable time scale.



Edited 1 time(s). Last edit at 04/26/2016 06:12PM by bobafetthotmail.

That sounds alot like same isse I got =/

Fotex Wrote:
-------------------------------------------------------
> That sounds alot like same isse I got =/

I'm not sure. They reported the Intel NIC hung, but it did not freeze the LAN for other non-Intel boxes. So your problem might be different. As I suggested, you can run top and log to a file. When this problem occurs in the future, top log might point to a buggy process that hogged the network. Or the lack of that symptom would indicate the NIC driver itself has caused it.

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)

> They reported the Intel NIC hung, but it did not freeze the LAN for other non-Intel boxes.

I'm ready to bet that the Phoronix guy is using pro networking gear like say a pro switch for the very least (he has 20 or 30 PCs in his basement for tests and benches), while Fotex is likely using a consumer-grade device.

It's not the first time a consumer-grade router or switch acts strange on me, I always flash them with OpenWRT or dd-wrt for a reason.

I did switch the router first time this happened, but That router was already flashed With dd-wrt.

Im currently running the 5sec top log.

- fotex