LinuxQuestions.org

LinuxQuestions.org (/questions/)
-   Linux - Networking (https://www.linuxquestions.org/questions/linux-networking-3/)
-   -   Internet works about 5% of the time (Slackware 12.1) (https://www.linuxquestions.org/questions/linux-networking-3/internet-works-about-5-of-the-time-slackware-12-1-a-672606/)

Dachy 09-26-2008 06:27 PM

Internet works about 5% of the time (Slackware 12.1)
 
I just installed Slackware 12.1 on a DELL PowerEdge 1300. It has a brand new USRobotics Gigabit Ethernet PCI Adapter 10/100/1000mbps. I put this in there just in case it was an hardware issue but I still get the same results.

This box is a central logging server running Nagios and pulling information from Windows XP PCs running NSClient.

On Nagios the host details show those PCs as being down sometimes, then up. 0% packet loss to 100%.

Any ideas?

unSpawn 09-27-2008 04:05 AM

What tests are used to determine "up" status? Wrt tolerance, what's the timeout for state change for those tests? Wrt granularity, at what interval are the tests run? Is there a pattern in up/down state changes? If you manually monitor one of these hosts for a prolonged period with equivalent commandline tools (Tcptraceroute, Hping?) does that support the idea of host or network connectivity problems? If you monitor the Nagios hosts for a prolonged period, does it show ethernet link state problems (ethtool, miidiag or equiv), network connectivity problems, Nagios process latency or other problems due to machine load?

Dachy 09-29-2008 10:13 AM

Quote:

Originally Posted by unSpawn (Post 3293188)
What tests are used to determine "up" status? Wrt tolerance, what's the timeout for state change for those tests? Wrt granularity, at what interval are the tests run? Is there a pattern in up/down state changes? If you manually monitor one of these hosts for a prolonged period with equivalent commandline tools (Tcptraceroute, Hping?) does that support the idea of host or network connectivity problems? If you monitor the Nagios hosts for a prolonged period, does it show ethernet link state problems (ethtool, miidiag or equiv), network connectivity problems, Nagios process latency or other problems due to machine load?

this is a bit over my head.. as in, I'm not very familiar with the process of monitoring the hosts using the tools you mentioned or what WRT is.

How exactly would I be able to test for wrt tolerance and granularity?

Thank you in advance

unSpawn 09-30-2008 01:50 PM

I'll skip the explanation if you don't mind and suggest you manually run a continuous ping from the Nagios box and another machine to one of the troublesome machines. If errors occur it should be easier to determine what goes on where. TIA

Dachy 09-30-2008 06:42 PM

I've noticed this when i boot up:

dnsmasq: failed to create listening socket: address already in use

when i did the continuous ping, at one point the hosts would come back up and internet would be accessible, then they would go down and of course internet would not work.

unSpawn 10-01-2008 04:28 PM

I hope somebody would be willing to jump in right now, because I don't think I am the right person to help you further. Please accept my apologies.

Dachy 10-01-2008 04:31 PM

Quote:

Originally Posted by unSpawn (Post 3297424)
I hope somebody would be willing to jump in right now, because I don't think I am the right person to help you further. Please accept my apologies.

I appreciate the effort, I know its frustrating working with someone who doesn't have a clue.

As I look into it more I'm finding out more. I was trying to set this box up with a static ip and can't get that to work properly, so for now I'm using DHCP until I can figure it out on a spare pc.

thanks again unSpawn for your patience.

ernst 10-02-2008 06:38 AM

Just a guess.
Are you aware of tcp_window_scaling ? See:
http://lwn.net/Articles/92727/

unSpawn 10-07-2008 01:50 AM

So. Did this get fixed Dachy?


All times are GMT -5. The time now is 02:17 PM.