Multicast interface vs udp-sender parameters



  • Hi!

    We have and issue in which a partial problem of multicast is detected which can be solved if multicast can be configured somehow. In this setup we have an interface called p4p1 (in old days it would be eth0). During fog setup it asks if i want to change default network interfacename, and i answered yes, p4p1 will be the one. Some places it still used eth0 in web gui, so i changed them (i dont remember where, i found it when i wanted to find else).

    The original case came to light cos I set up one physical interface with two ip address. One for maintenance, one for production. As we debugged the multicast problem it came to light that sender waits for fog one ip, receiver tries to get to the other one. On debug web page there is a place where upd-receiver is configured with rcv-address and if I write here the “maintenance ip” multicast debug successfully done.

    So, I guess that if I can tell somehow the the udp-sender/udp-receiver duo those parameters which we used in debug sessions then it can work somehow.

    What do you think?



  • @Wayne-Workman

    Er… during your post I was trying to find a better way to describe why was 2 different post, and tried to state better what I wanted. Sorry again for being a bit confusing. Pls read what I wrote now. Anyway, p4p1 thing you suggested is a bit more dangerous than you may wanted :D It is like killing a moskito with gunship, full power. It will kill, but… well, a bit more damage than wanted. :)

    My problem maybe originated that some things can be configured on the web ui, but sometimes config information is kept in different location which is ok if I dont want to reconfigure “all things” after setup. It is hard to be done if I dont know where to make changes (ip addresses are located in many places, but in config files duplicated, cannot be modified from webui. This makes things harder for debugging.



  • Hi!

    Well, I am sorry for confusing you, but last 2 topics i created was leading to totally different way, even if we have one in common, the fail of multicast. Let me be a bit more explanative.

    I still have multicast issues, which we trying to clear a bit to let us be more on to solve the true problem. The setup is the same server:

    • ubuntu 14.04 lts
    • network interface, and on same interface virtual interface. first one is mantenance, other is for actual fogging.
    • clean install, no previous life of it, only 1-2 images from old days, later converted to new format btw.

    Before multicast try we noticed the “storagenode not found” thing, which indicated some problem (first topic). We have working unicasts.

    As we tried to debug, we found a partially working stage in which we investigated with debug steps and found that upd-sender manual start on server and udp-receiver on client we can make working version, but we have a problem, server and client are “listening on different direction” or so. If we have set the optional parameter on receiver about udp address (–mcast-rdv-address fogserverIP) it works. But, as all time you can find a “but”, we have to set this address to the “maintenance address” of the server, not the address of the web interface, tftp ip, etc.

    Then we moved on. Killed the “2 interface address setup” (here we found the changing the server ip is a lot harder than the 0.32 version; an update of the wiki is a must on this topic). Now we have only the p4p1 named network interface, have only one ip address, unicast working, multicast not working. Here I forked my post as possibly this is not same issue.

    • question1: with working unicast what does “no storage node found” actully mean?
    • question2: is it possible to set parameters for multicast? like i mentioned above, mcast-rdv-address, or others

    So, I apologise for virtually repost, but if we find any solution it would be easier to find later by others. Or at least it was my intention by forking the question.


  • Moderator

    @Foglalt said:

    We have and issue in which a partial problem of multicast is detected which can be solved if multicast can be configured somehow. In this setup we have an interface called p4p1 (in old days it would be eth0). During fog setup it asks if i want to change default network interfacename, and i answered yes, p4p1 will be the one. Some places it still used eth0 in web gui, so i changed them (i dont remember where, i found it when i wanted to find else).

    Complicated…

    Here’s how you would replace EVERY SINGLE INSTANCE of eth0 with p4p1 in EVERY SINGLE FILE in the OS… Very scary… use with discretion.

    grep -rl 'eth0' / | xargs sed -i 's/eth0/p4p1/g'
    

    Here’s the reference I used:
    http://vasir.net/blog/ubuntu/replace_string_in_multiple_files

    And I agree with @Tom-Elliott , You haven’t clearly stated a problem and I’m not sure what you’re asking…


  • Senior Developer

    Can you clarify what the problem is?

    I’m not understanding. You have multiple posts but keep creating new posts about things. It’s very hard to discern what it is you’re looking for.

    The other post where it says StorageNode is not the master, is this still the case? Changing the interface of which is being searched may help, but it is no guarantee because I believe the 1.2.0 just scanned all IPs and tries to guess based on those whether or not it is the master. Of course I could be wrong, but it’s what i understand.

    Is there any possibility of installing SVN/Trunk (when Sourceforge finally comes back up)?



  • At boot process of the system it is named. System logs, ifconfig, ip command. All says it is so.


  • Senior Developer

    How do you know the interface’s name is p4p1?


Log in to reply
 

368
Online

38983
Users

10712
Topics

101680
Posts

Looks like your connection to FOG Project was lost, please wait while we try to reconnect.