+ Reply to Thread
Page 1 of 2 12 LastLast
Results 1 to 10 of 13
  1. #1
    Join Date
    Apr 2004
    Location
    Arkansas, USA
    Posts
    867
    Rep Power
    11

    Default Clocksource tsc unstable

    I'm running 1.5.16.7 server. It is running on a Soekris net5501 board. I've noticed drop-outs on my ethernet port on this router. I have installed several of these in the past, with older firmware and not had a problem.

    What I'm seeing is, every 15-20 mins, I get a message in the log saying...

    kernel: Clocksource tsc unstable (delta = 3389103963 ns)
    kernel: Switching to clocksource pit
    kernel: e100 0000:01:04.0: eth0: NIC Link is Up 100 Mbps Full Duplex

    Then about 20 mins later, I get

    kernel: e100 0000:01:04.0: eth0: NIC Link is Up 100 Mbps Full Duplex

    Also, as I mentioned in another post, the ISC-DHCP doesn't work. It says can't write to database or something.. not sure if that is related.

    Thanks,
    Roger

  2. #2
    Join Date
    Apr 2004
    Location
    Arkansas, USA
    Posts
    867
    Rep Power
    11

    Default

    Furthermore... the ethernet port in question is an add-on card I got off of ebay. So I tried moving the IP and ethernet cable from that port to one of the on-board ports, and I could not for the life of me get OLSR to communicate. (Yes I have OLSR set up to run on all 6 ethernet ports in the config). I moved it back to the original eth0 port, and it started working again.

    Thanks,
    Roger

  3. #3
    Join Date
    Oct 2002
    Location
    Pattaya, Thailand
    Posts
    11,778
    Rep Power
    10

    Default

    It is probably an ARP cache issue with other units on the network. It takes time for them to clear their cache.

    Quote Originally Posted by Beebe View Post
    Furthermore... the ethernet port in question is an add-on card I got off of ebay. So I tried moving the IP and ethernet cable from that port to one of the on-board ports, and I could not for the life of me get OLSR to communicate. (Yes I have OLSR set up to run on all 6 ethernet ports in the config). I moved it back to the original eth0 port, and it started working again.

    Thanks,
    Roger

  4. #4
    Join Date
    Oct 2002
    Location
    Pattaya, Thailand
    Posts
    11,778
    Rep Power
    10

    Default

    My conclusion would be that the hardware clock to the Soekris is not regular and the kernel switches on its own to try and correct it. We'll check the Server and see if there is a missing file or directory. ISDHCP is pretty brain dead about missing files, when instead of simply creating an empty on it simply refuses to start. It works great as long as things are proper and maybe we messed up with the layout.

    Quote Originally Posted by Beebe View Post
    I'm running 1.5.16.7 server. It is running on a Soekris net5501 board. I've noticed drop-outs on my ethernet port on this router. I have installed several of these in the past, with older firmware and not had a problem.

    What I'm seeing is, every 15-20 mins, I get a message in the log saying...

    kernel: Clocksource tsc unstable (delta = 3389103963 ns)
    kernel: Switching to clocksource pit
    kernel: e100 0000:01:04.0: eth0: NIC Link is Up 100 Mbps Full Duplex

    Then about 20 mins later, I get

    kernel: e100 0000:01:04.0: eth0: NIC Link is Up 100 Mbps Full Duplex

    Also, as I mentioned in another post, the ISC-DHCP doesn't work. It says can't write to database or something.. not sure if that is related.

    Thanks,
    Roger

  5. #5
    Join Date
    Oct 2002
    Location
    Pattaya, Thailand
    Posts
    11,778
    Rep Power
    10

    Default

    Tony has confirmed the ISDHCP issue will be resolved. It is a missing leases file and scripting can fix it.

  6. #6
    Join Date
    Apr 2004
    Location
    Arkansas, USA
    Posts
    867
    Rep Power
    11

    Default

    Brilliant, thanks!

  7. #7
    Join Date
    Apr 2004
    Location
    Arkansas, USA
    Posts
    867
    Rep Power
    11

    Default

    Quote Originally Posted by lonnie View Post
    It is probably an ARP cache issue with other units on the network. It takes time for them to clear their cache.
    Isn't the ARP cache cleared by an activate changes? Because I did that on the neighbouring units.

    Thanks,
    Roger

  8. #8
    Join Date
    Oct 2002
    Location
    Pattaya, Thailand
    Posts
    11,778
    Rep Power
    10

    Default

    The ARP cache of all the connected machines is the issue and you an reboot the unit all day long with no effect on the other machines who associate the MAC address to an IP. They either need time for a proper timeout or they need to be cleared manually. Remember the issues with Windows machines when you would try several units all at 192.168.1.1? The second one would not be talked to until you cleared the Windows ARP table.

  9. #9
    Join Date
    Apr 2004
    Location
    Arkansas, USA
    Posts
    867
    Rep Power
    11

    Default

    It's the connected machines I am asking about. Does an activate changes clear the arp cache? Because I DID activate on the neighbouring machines.

  10. #10
    Join Date
    Oct 2002
    Location
    Pattaya, Thailand
    Posts
    11,778
    Rep Power
    10

    Default

    Activate only touches what has been changed. If you made no changes to the connected machine Ethernets then they will probably not clear their ARP tables.

    Are you certain you did not have any static routes in the affected system?

Similar Threads

  1. Full Duplex Causes Quality to drop, link unstable
    By cephlon in forum WAR Board -- Wireless Advanced Router
    Replies: 5
    Last Post: 06-11-2008, 05:13 PM
  2. Unstable Core Router
    By Beebe in forum V3 Topics
    Replies: 14
    Last Post: 08-07-2007, 08:51 PM
  3. Unstable association with WPA-PSK
    By williamm in forum Support
    Replies: 21
    Last Post: 12-04-2006, 02:13 PM

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts