Multicast to 10 PC creates two upd-sender processes



  • Hi all,
    just a question: i created a group with 10 pc and I created a multitask deploy job for that group.
    Fog created two different row in the multicastSessions table one with 7 msClients and one with 3 msClients.
    This is slowing down the whole deploy process (more network traffic, server I/O contention because of the two udp-sender processes).
    Is this behaviour normal? I supposed to have just one multicastSessions with 10 clients.

    I’m using fog svn 3559 on Centos 6.6.



  • @Tom-Elliott said:

    @brizio Can you clear the multicastSessions, multicastSessionsAssoc, and tasks tables?

    The tasks I only want you to clear the multicast jobs.

    A mysql statement to do these things would be:

    TRUNCATE TABLE multicastSessions;
    TRUNCATE TABLE multicastSessionsAssoc;
    DELETE FROM tasks WHERE taskTypeID=8;
    

    My guess, from an earlier revision, a multicast session didn’t clear properly. Because of the non-clearing, a few hosts linked to the wrong Job and the others, who had no prior setting stored in the DB, created their own session.

    Again, this is just a guess, as I really don’t know what actually caused this to happen.

    I’m working on a “Troubleshoot Multicast” article that has commands more friendly to Red-Hat based distros… aparently the truncate commands don’t work in Fedora 22… so… Here you go:

    https://wiki.fogproject.org/wiki/index.php/Troubleshoot_Downloading_-_Multicast



  • Right, i’ll clean the DB then. Thank you.


  • Senior Developer

    @brizio Can you clear the multicastSessions, multicastSessionsAssoc, and tasks tables?

    The tasks I only want you to clear the multicast jobs.

    A mysql statement to do these things would be:

    TRUNCATE TABLE multicastSessions;
    TRUNCATE TABLE multicastSessionsAssoc;
    DELETE FROM tasks WHERE taskTypeID=8;
    

    My guess, from an earlier revision, a multicast session didn’t clear properly. Because of the non-clearing, a few hosts linked to the wrong Job and the others, who had no prior setting stored in the DB, created their own session.

    Again, this is just a guess, as I really don’t know what actually caused this to happen.



  • It seems also that the upd-sender process started with
    min receivers=5 for the first session (the one set with msClients=7)
    min receivers=3 for the second session (the one set with msClients=3)

    that means that 2 PC were not depoloyed: in fact they are still waiting on the first PartClone screen.



  • Hi Tom,
    thank you for the quick reply.
    Yes, all the 10 PC in the same group are set with the same Image Id (ID: 9).
    I can see the same ID also in the multicastSessions table, column msImage.


  • Senior Developer

    I guess we need details.

    Are all systems in the group set with the same image id?


 

570
Online

5.4k
Users

12.6k
Topics

118.7k
Posts