Where is Zapata, Dahdi is not being nice Plz help

Discussion in 'General' started by rollinsolo, Apr 16, 2009.

  1. rollinsolo

    Joined:
    Feb 11, 2009
    Messages:
    279
    Likes Received:
    0
    Does anyone have a good fix for this yet, I have to use Dahdi if Zapata is gone.

    [Apr 15 21:00:36] VERBOSE[4944] logger.c: Asterisk Event Logger Started /var/log /asterisk/event_log
    [Apr 15 21:00:36] ERROR[4944] asterisk.c: Asterisk has detected a problem with y our DAHDI configuration and will shutdown for your protection. You have options :
    1. You only have to compile DAHDI support into Asterisk if you need it. One option is to recompile without DAHDI support.
    2. You only have to load DAHDI drivers if you want to take advantage of DAHDI services. One option is to unload DAHDI modules if you don't need them.
    3. If you need Zaptel services, you must correctly configure DAHDI.
    [Apr 15 21:00:40] VERBOSE[4969] logger.c: Asterisk Event Logger Started /var/log /asterisk/event_log
    [Apr 15 21:00:40] ERROR[4969] asterisk.c: Asterisk has detected a problem with y our DAHDI configuration and will shutdown for your protection. You have options :
    1. You only have to compile DAHDI support into Asterisk if you need it. One option is to recompile without DAHDI support.
    2. You only have to load DAHDI drivers if you want to take advantage of DAHDI services. One option is to unload DAHDI modules if you don't need them.
    3. If you need Zaptel services, you must correctly configure DAHDI.
    [root@elastix ~]#
     
  2. rollinsolo

    Joined:
    Feb 11, 2009
    Messages:
    279
    Likes Received:
    0
    Sorry little more info I am using a Rhino R1/T1 card and everything was working good before I put the card in. Hmmmm.
     
  3. rollinsolo

    Joined:
    Feb 11, 2009
    Messages:
    279
    Likes Received:
    0
    Can anyone show me a sample file for what Dahdi is supposed to look like in dahdi_channels.conf for an ie. 12 channel PRI. Can I just copy the settings from an older zapata_channels.conf. Or ofcourse I can load 1.3 and move on with my life, but if dahdi is the future I would love to learn it, I spent a couple of hours looking through the forums last night and cannot find any examples for this.
     
  4. dicko

    Joined:
    Oct 24, 2008
    Messages:
    4,099
    Likes Received:
    0
  5. rollinsolo

    Joined:
    Feb 11, 2009
    Messages:
    279
    Likes Received:
    0
    Ok after playing with Dahdi for a while, I am still a little confused at to which file I will edit to configure a PRI, I would think dahdi_channles.conf is where i would add, I have a 12 Channel PRI, so at the bottom I have changed 1-2 to 1-12. But in Elastix hardware detection it does not show any Channels, all I get is

    Span # 1: DAHDI _DUMMY/1 "DAHDI_DUMMY/1 (source: Linux26) 1" (MASTER)
    Ports not Founds


    context=from-pstn
    signalling=pri_cpe
    faxdetect=incoming
    usecallerid=yes
    echocancel=yes
    echocancelwhenbridged=no
    echotraining=800
    group=0
    channel=1-2



    Has anyone else successfully configured a PRI to work with Elastix 1.5 and can I see your different .conf files to try to replicate. Thanks.
     
  6. ramoncio

    Joined:
    May 12, 2010
    Messages:
    1,663
    Likes Received:
    0
    Try this in the bash console:

    amportal stop
    dahdi_cfg -v
    /etc/init.d/dahdi restart
    dahdi_genconf
    amportal start
    lsdahdi
     
  7. dicko

    Joined:
    Oct 24, 2008
    Messages:
    4,099
    Likes Received:
    0
    apparently the hardware detection did not work.

    I never rely on "auto detection" it can never be guaranteed, so manually set up chan_dahdi.conf as per the voip-info site. I wouldn't be able to provide anything useful as I've never heard of a 12 channel pri, are you sure you don't have a traditional "fractional" T1?

    All PRI have 24 channels (in the US), 1 for signaling and the others (not all need to be available) for calls
     
  8. ramoncio

    Joined:
    May 12, 2010
    Messages:
    1,663
    Likes Received:
    0
    When you configure a card there are two parts.
    First part: you install the linux module, configure the /etc/dahdi/system.conf and start the dahdi service with /etc/init.d/dahdi restart. This is like installing a driver in Windows.

    This is an example of how I configured this file for my bri card:

    Code:
    # Span 1: B4/0/1 "B4XXP (PCI) Card 0 Span 1" (MASTER) AMI/CCS 
    span=1,1,0,ccs,ami
    # termtype: te
    bchan=1-2
    hardhdlc=3
    echocanceller=oslec,1-2
    
    # Span 2: B4/0/2 "B4XXP (PCI) Card 0 Span 2" AMI/CCS RED
    span=2,2,0,ccs,ami
    # termtype: te
    bchan=4-5
    hardhdlc=6
    echocanceller=oslec,4-5
    
    
    # Global data
    
    loadzone	= es
    defaultzone	= es
    
    bchan are the data channels, and hardhdlc is for the signalling channel.

    Elastix includes most common modules, so you just have to configure your system.conf and restart dahdi.
    You should see your card loaded ok with lsdahdi or dahdi_hardware and you should see the module loading in the linux kenel with dmesg.
    This can also be done automatically with the command dahdi_cfg -v

    Second part: You have to configure asterisk to work with your hardware. This is done by chan_dahdi.
    In /etc/asterisk/chan_dahdi.conf you configure the global dahdi settings. At the end of this file you must have this line that includes the channels:

    #include dahdi-channels.conf

    Finally, in /etc/asterisk/dahdi-channels.conf you configure your channels specific settings, as signalling, etc. You can also do it automatically with dahdi_genconf. And after that, you restart asterisk to reload the new configuration, and it should work.
     
  9. rollinsolo

    Joined:
    Feb 11, 2009
    Messages:
    279
    Likes Received:
    0
     
  10. rollinsolo

    Joined:
    Feb 11, 2009
    Messages:
    279
    Likes Received:
    0
    Dicko these are dynamic T1's so I could sell someone a 12 Channel PRI, 5 analog on to a 66 Block, and since its dynamic, when no one is one the phone they will have a full 1.5 pipe for data. This is not TDM and with 1.3 we had to be right on with the channels or calls were screwy, I think its because incoming starts with Zap1/1 and outbound started at the last b channels so ie zap1/17/. So I guess If I left it open for a full 24 it would waste time searching for channels that aren't there until it found 17 for outbound. So with the company I sell T1's through the D channel is always on 24, not dictated like you can with TDM.
     
  11. dicko

    Joined:
    Oct 24, 2008
    Messages:
    4,099
    Likes Received:
    0
    My understanding is that PRI's are a version of TDM where the inband signalling found in traditional T1's which are designed to fit in a 1.54Mb (US T1, (T for Telephony) DS1)) are delegated to the D channel. if you use a TDM over IP solution, I believe that these timing signals (and channels) are reconstituted at the far end hardware to a a full 24 channel (T1) TDM signal for interface with the customer CSU. When I do this (I use RAD hardware) the bandwidth is dynamic indeed between the boxes, but the CPE handoff to either T1 or FXO is full DS0 (64k ulaw) channels.
    Further, as the call setup/pulldown is done with reference to the D Channel you should not need to use the ZAP/G0 construct as channels are assigned with pre-cognition, hence there is no chance of "traditional" glare where winks collide (ZAP/g0 should be fine) yes limit the B channels to suit bandwidth constraints, for good practice use pri_net signalling at the asterisk box (pri_CPE at the customer premise if it's a PBX) (make sure the TDM over IP boxes agree with this timing) and the switchtype = national
     
  12. dicko

    Joined:
    Oct 24, 2008
    Messages:
    4,099
    Likes Received:
    0
    If not TDM over IP, and you are actually riding on a dynamic 2-pair T1 (TDM) to the CPE then the hardware T1 multiplexing used by your company for the dynamic aspect should take care of your interface to the provisioning hardware and the CPE demultiplexing will provide a full 24 channel PRI interface at the CPE (rhino) , extra to the IP services your company hands off for data.

    To summarize:

    PRI's are by definition TDM (time domain multiplexed). they have 24 channels of 64kbs (in the US) they have D (data channels) for signalling and B (bearer) channels for voice calls, by convention D channel is in time slot 24 in the US.
    They can be virtualized, compressed and multiplexed, but end to end they are absolutely 1.54Mbs channels, Thus they must follow all the timing/framing constraints imposed by the "other end" required by T(E)1's
    They use ISDN signalling (which is of course tautological) in one or more D channels to efficiently manage the B channels.
    Timing, framing and line coding is crucial for correct operation (the underlying T1 protocol is synchronous) There should only be one master timing source, this is normally accepted from the "network", if you are the only T1 channel on a node then be master, if you are co-mingled with other sources of clock then "caveat".

    (There is a whole bunch more about PRI parameters that might cause your call to be rejected (in your other thread) depending on all sorts of things. Whilst I appreciate your efforts to be a "telco", with that goes the precondition that you be "five nines", Go for it , knowledge is king!)

    hint: from asterisk CLI, try "pri debug"
     
  13. ramoncio

    Joined:
    May 12, 2010
    Messages:
    1,663
    Likes Received:
    0
    Leave the echocanceller=oslec,1-12 line, as oslec is the software echo canceller integrated in Elastix, and it does a really good job.
     
  14. rollinsolo

    Joined:
    Feb 11, 2009
    Messages:
    279
    Likes Received:
    0
    Ok so it should look like this, I think for what I have read for a us pri it should read dchan instead of the one you use. will try it with both.

    span=1,1,0,esf,b8zs
    # termtype: te
    bchan=1-12
    dchan=24
    echocanceller=oslec,1-12 line

    # Global data

    loadzone = us
    defaultzone = us
     
  15. dicko

    Joined:
    Oct 24, 2008
    Messages:
    4,099
    Likes Received:
    0
    I refer you to this further reference at voip-info.org

    http://www.voip-info.org/wiki/view/Asterisk+PRI

    (and all it's sub references)


    It might give you some insight as to how PRI's really work and also some insight as to the nature of the T1's that they ride on, as you are apparrently acting as the provider then spend some extra attention to the concept of master clocking. and choice of switchtype signaling. You have assumed the position of a system engineer, so you really need to raise yourself above the level of "tech."
     
  16. rollinsolo

    Joined:
    Feb 11, 2009
    Messages:
    279
    Likes Received:
    0
    Ok, so this would address the new almighty, going to be with us for a long time so we must conform Dahdi, but 1.5 is supposed to be stable, and I don't mean this as complaining but when you load 1.3 after raid there is no issues, it just goes straight to the root password recognizes more NIC cards right away no IRQ issues, and zapata is ready to go from hitting hardware detection button and then configuring once .conf file to bring in the pri, Im just curious what the issues with 1.5 are and why it is called stable. Compared to 1.3 there are many hurdles to overcome just to move on to configuring the system. I am not complaining at all because I have tried them all and Elastix is the best, its great to see new releases are rolling out and people are interested in their product and the future of it. I just think it should be called Beta still but then again Im a newb. I have to pop my 3 Com card out between my two PCI slots until it recognizes it and puts it on eth1 and not eth0. I just think I have a lot more to learn before I move to 1.5.

    I want to give it one more shot before I put it on the shelf for another month, meaning putting it in production, when I load it for the first time and it does not recognize my card, and I cannot go online, how would I load the drivers for 3com and make sure it is assigned eth0, this is automatic with 1.3.
     
  17. dicko

    Joined:
    Oct 24, 2008
    Messages:
    4,099
    Likes Received:
    0
    I understand your frustrations, but please don't blame eleastix, it sits on centos and any kernel must sit on the hardware platform you deploy, vagaries in interrupt sharing and bios settings (particularly the so called plug and play turn it off!!) means that until you come to terms with how linux has evolved and streamlined it's hardware detection and how udev handles it. The good thing is that ultimately you will have a more stable and faster system, (after the tears of course) basically a couple of udev rules will stop that frustration, unfortunately there are no general recipes here, just lots of specific solutions for hardware mixtures (google is your friend).

    good luck and hang in!!
     
  18. rollinsolo

    Joined:
    Feb 11, 2009
    Messages:
    279
    Likes Received:
    0
    I am not going anywhere, like I said in some other forum I am very stubborn and will continue to learn, adapt, and share. Life is just so busy for me right now it is so hard to take 30 steps back to take thousands forward, I think the smoke will clear after I go down to one job at the end of the month. Hopefully by then I can take a 500 foot view of Centos and develop better skills on the underlying architecture that elastix interfaces with. I guess I really don't want it to be too easy, I like challenges, but I hate deadlines. That is what I am faced with now. So thanks for your help.
     

Share This Page