View previous topic :: View next topic |
Author |
Message |
Raniz l33t
Joined: 13 Sep 2003 Posts: 967 Location: Varberg, Sweden
|
Posted: Mon Apr 20, 2009 1:32 pm Post subject: Server drops all packages from time to time |
|
|
I've recently put up a server in the cellar here, everything is working as it should except that it gets unresponsive from time to time.
At random intervalls it will just drop the network connection for a while and the return as nothing happened. I have no messages in /var/log/messages about this so I haven't got a clue about what's causing this.
Packages transmitted to the server is just dropped.
Code: | $ ping my_server
PING my_server (x.x.x.x) 56(84) bytes of data.
^C
--- my_server ping statistics ---
7 packets transmitted, 0 received, 100% packet loss, time 6004ms |
When I set up the server I did it at my place behind my own router and everything worked fine here, this started when I moved it to the basement. The only things that differs between the basement and my home LAN is the IP configuration, the speed (I've got 100mbps and now it's connected to a 1000mbps line) and the network switch that it's attached to.
Any ideas about what can cause this behaviour? |
|
Back to top |
|
|
danomac l33t
Joined: 06 Nov 2004 Posts: 881 Location: Vancouver, BC
|
Posted: Mon Apr 20, 2009 4:55 pm Post subject: |
|
|
What kernel are you running? 2.6.29 had issues with some network modules that would do that. You might want to take a look in the 2.6.29 thread. |
|
Back to top |
|
|
Raniz l33t
Joined: 13 Sep 2003 Posts: 967 Location: Varberg, Sweden
|
Posted: Mon Apr 20, 2009 10:21 pm Post subject: |
|
|
Code: | $ uname -rv
2.6.26-hardened-r9 #3 SMP Sun Feb 22 18:43:59 CET 2009 |
So it's not 2.6.29 related :/
But I'll try upgrading the kernel. |
|
Back to top |
|
|
poly_poly-man Advocate
Joined: 06 Dec 2006 Posts: 2477 Location: RIT, NY, US
|
Posted: Mon Apr 20, 2009 10:43 pm Post subject: |
|
|
#1: check the network itself - are other packets making it through to that area of the network?
#2: what says /var/log/messages about that situation?
also, check your power and temperature/humidity - could be messing with the computer. _________________ iVBORw0KGgoAAAANSUhEUgAAA
avatar: new version of logo - see topic 838248. Potentially still a WiP. |
|
Back to top |
|
|
ScarletPimpFromHell n00b
Joined: 15 May 2007 Posts: 32 Location: Sydney
|
Posted: Mon Apr 20, 2009 10:56 pm Post subject: |
|
|
Check the speed and duplex seetings of the switch and the server. Make sure they match.
The most common cause of intermittant network failure is multiple collisions when the server and/or switch is operating in half duplex mode.
After 16 successive collisions the switch will most likely perform a software reset on the interface, preventing all traffic flow during the reset. If this is occuring you should see the led above/below the server connection turn amber. I suggest you start a large file transfer and then run down to the basement and watch to see if the LED changes colour. |
|
Back to top |
|
|
Raniz l33t
Joined: 13 Sep 2003 Posts: 967 Location: Varberg, Sweden
|
Posted: Mon Apr 20, 2009 11:02 pm Post subject: |
|
|
poly_poly-man wrote: | #1: check the network itself - are other packets making it through to that area of the network? |
The server next to it is working fine.
Quote: |
#2: what says /var/log/messages about that situation? |
Only thing in the logs about networking is this:
Code: | tg3: eth0: Link is up at 1000 Mbps, full duplex.
tg3: eth0: Flow control is off for TX and off for RX. |
Quote: | also, check your power and temperature/humidity - could be messing with the computer. |
There are four other servers in the same room that are working fine, so that shouldn't be an issue.
Last edited by Raniz on Mon Apr 20, 2009 11:06 pm; edited 2 times in total |
|
Back to top |
|
|
Raniz l33t
Joined: 13 Sep 2003 Posts: 967 Location: Varberg, Sweden
|
Posted: Mon Apr 20, 2009 11:04 pm Post subject: |
|
|
ScarletPimpFromHell wrote: | Check the speed and duplex seetings of the switch and the server. Make sure they match.
The most common cause of intermittant network failure is multiple collisions when the server and/or switch is operating in half duplex mode.
After 16 successive collisions the switch will most likely perform a software reset on the interface, preventing all traffic flow during the reset. If this is occuring you should see the led above/below the server connection turn amber. I suggest you start a large file transfer and then run down to the basement and watch to see if the LED changes colour. |
Both the server and the switch are running in gigabit mode, fairly certain that both are using full duplex. I'll start a large transfer and run down to check it tomorrow |
|
Back to top |
|
|
ScarletPimpFromHell n00b
Joined: 15 May 2007 Posts: 32 Location: Sydney
|
Posted: Mon Apr 20, 2009 11:48 pm Post subject: |
|
|
Hmm, gigabit over copper = short distances. How long is the cable drop from switch to server ?
Just had another thought. The 802.3ab standard uses all four pairs of copper. You might want to insure you are using HIGH grade copper cable. Ie Category 5e at the very least, category 6 to be sure.
Also from memory I think the recommended maximum distance is about 10 meters with cat 5e and 90 meters with cat 6. I'll have to go and look this up though. I don't remember the exact measurements. |
|
Back to top |
|
|
Raniz l33t
Joined: 13 Sep 2003 Posts: 967 Location: Varberg, Sweden
|
Posted: Tue Apr 21, 2009 10:33 am Post subject: |
|
|
I've tried with both 3m and 1m cables, I'm unsure if they were cat 5 or 5E though.
I've moved the server to a 100mbit switch now to see if that's the issue. |
|
Back to top |
|
|
|