TFTP Server Stopped Responding

  • This topic is empty.
Viewing 3 posts - 1 through 3 (of 3 total)
  • Author
    Posts
  • #893
    Joe
    Member
    • Total Post: 96
    • Back Stage Pass
    • ★★★★

    I’ve been having some problems PXE booting and I’ve found that the TFTP service stops responding to connection requests.

    Server: Windows 2003
    App: Wyse Device Manager 4.5.1
    DB: SQL Server 2000

    I first noticed the issue when the devices stopped PXE booting. They would receive the job, prompt the user to Update, and then reboot. During PXE boot, everything looked fine but it would always go to local boot instead of network boot.

    I fired up Ethereal and logged the PXE process. I saw the DHCP request, I saw the Acks… everything looked fine.

    I fired up netstat to verify that the standard service was listening on port 69 (tftp), and it was…

    I found that if I stopped the Rapport4 service, then started it again, I could deploy one image.

    Curious, I duplicate the issue using the TFTP.EXE command line. I dumped a text file into the TFTP directory and use the command line to download it. At first, I received an error message telling me the connection timed out. I restarted the RAPPORT4 service, and it worked perfectly from the command line.

    To remove the possibility of a network problem, I decided to try the TFTP.EXE test on the WDM server, and it worked identical to a remote machine — the first time you connect with TFTP, it works fine. After that, it stops accepting connections and you receive a timed out error on the client.

    Interesting enough, there are no TFTP logs when this happens. It’s like the TFTP service just stops working. There are no events in the event logs that correspond to these the times I’m trying to TFTP, either.

    The *ONLY* change that I have made recently is that we moved from one database server to another, which is something I’ve done before and feel comfortable that I performed correctly (just changed the DB instance name in the registry).

    Anyone have any Ideas?

    #11550
    karaziel
    Member
    • Total Post: 74
    • Back Stage Pass
    • ★★★★

    If the only thing you changed was the registry, then you have not changed enough to do this correctly.

    Update the INSTALL table of the RapportDB to reflect the new SQL servername
    Clear the SERVER table so the services can checkin properly (common issue if you backed up DB while Services were connected)
    Check for entries in the COMMAND and COMMANDARG tables, and delete if present

    Its possible your PXE service was unable to connect to the DB correctly, or that moving your DB has left orphaned updates in the Command table from previous jobs which is adversely affecting your new updates.

    I have a habit of breaking and fixing things, and these are tricks I’ve learned.

    try these changes and see if it clears things up for you.

    If you aren’t seeing enough info in the logs, try setting them to DEBUG.
    If the logs are way behind, try clearing the SystemLog table (its where the logs are stored) TRUNCATE TABLE SYSTEMLOG is the fastest method…Delete takes to long.

    Also, while you’re making changes, turn off the SNMP support in the WDM preferences, its just a resource hog. You’ll thank me for this one day.

    -k

    #11577
    Joe
    Member
    • Total Post: 96
    • Back Stage Pass
    • ★★★★

    Clear the SERVER table so the services can checkin properly (common issue if you backed up DB while Services were connected)

    Thats what did it. Clearing that table resolved the problem.

    Thanks for the info, I do appreciate it.

Viewing 3 posts - 1 through 3 (of 3 total)
  • You must be logged in to reply to this topic.