Elastix in major production environment

Discussion in 'General' started by JanKoster, Oct 18, 2009.

  1. JanKoster

    Joined:
    Oct 18, 2009
    Messages:
    29
    Likes Received:
    0
    Hi,
    I'd like to find out if the following is possible;
    (Elastix with A2Billing)
    Our company is planning to open a few "Telecoms" branches as an "Add-on" to our current business, being wireless internet provider.
    We foresee to have, per branch, more than 2000 telco customers. We'll be getting E1 lines and link that to Elastix via 8 port PRI cards. Now this will only give 240 channels per card, so Q1 is:
    Can Elastix machines, each having such an 8 port PRI card, be "bound" together to use one accounting system (A2B), and can the machines be bound to one public IP number? (We're thinking of using a blade server, 16 blades). So, maybe the question to ask, prior to the question above, is wether Elastix can scale up using blade?
    I have read about call limitations when G729 is used (vs G711), and still have to decide how to approach this (bandwidth). Calls will be made majorily via landlines (E1), but SIP dialing will also be done on a smaller scale.
    Any help/info appreciated.
     
  2. dicko

    Joined:
    Oct 24, 2008
    Messages:
    4,099
    Likes Received:
    0
    Yes it's all possible. All your questions asked and answered here already. Use the search thingy.
     
  3. ramoncio

    Joined:
    May 12, 2010
    Messages:
    1,663
    Likes Received:
    0
    Hi JanKoster, wellcome to the forum.
    You should consider using different hardware for your installation.
    If you plan to deploy a high availability solution (you MUST) it is not good to use internal cards, you should have a look at gateways, astribanks or redphones. With internal cards you would need to duplicate telephony hardware.
    For the ha cluster you could use heartbeat+dbrd, there is a tutorial somewhere here, search the forum/wiki. Also you should consider using sip proxies (openser, kamailio) and media proxies if you also plan to use many sip extensions. This would also increase scalability and performance of your system.
    Anyway, you should really think about getting someone with highest linux and asterisk expertise to deploy the installation for you, as this should't be a system to learn with.
     
  4. rbo

    rbo

    Joined:
    Oct 20, 2009
    Messages:
    3
    Likes Received:
    0
    Hi Jan Koster,

    If your locations are in the Netherlands, you probably are using VPN's to connect the locations, if so you can connect several elastix-servers together and operate as one, getting redundancy as a bonus.
    With QoS on these vpn's you can connect elastix-servers via iax, works great with connections like Epacity from KPN.
     
  5. JanKoster

    Joined:
    Oct 18, 2009
    Messages:
    29
    Likes Received:
    0
    Ramoncio
    Thx for your advice. :)
    I've now bought the Astribank with Twinstar system.
    I've read up how-to, and configured two Dell 1850 servers with Elastix 1.6, updated, and then installed DRBD and frills for HA.
    I now have it working and failing over. Astribank (4X PRI & 24 FXO) connected to both, and registered on node 1. No problems.
    But - :dry:
    The twinstar setup I can not find. (Maybe because I'm too tired). I've searched the Internet for a "how-to", even this forum on "twinstar" gives no results.
    I did find some document(s) on what to check, and all check fine.
    What happens, is on failover (forced), the Elasix system fails over, but the Astribank does not register on node 2. All lights come on, and "hardware detection" does not pick it up. Obviously because the Twinstar unit did not switch USB port from "Main" to "Backup".
    Do you have any advice for me (other than setting fire to the ##$*&&^ unit?
    To Xorcom: UPDATE your instructions!!! :angry:
    Jan (South Africa)
    Thx in advance :S
     
  6. JanKoster

    Joined:
    Oct 18, 2009
    Messages:
    29
    Likes Received:
    0
    It seems I need a file called "twinstar", located in the /usr/share/twinstar/ folder, or even in one of the "sbin" folders? My Elastix 1.6 does not have that. This is the utility that 'talks' to the TwinStar, switches watchdog on/off, which in turn switches the USB ports. I see in the "twinstar_hook" script file it runs that "twinstar" tool.
    Anybody that might have this file, Please...... I need it very urgently.
    Thx
    Jan
     
  7. dicko

    Joined:
    Oct 24, 2008
    Messages:
    4,099
    Likes Received:
    0
    I believe you need to find and run the autonomously named script twinstar_setup to make it, it is a a perl script that has been part of asterisk for quite a while now, if it isn't on your machine already you should be able to get it from the digium svn server, did you ask xorcom, they are usually quite responsive and helpful.

    dicko
     
  8. JanKoster

    Joined:
    Oct 18, 2009
    Messages:
    29
    Likes Received:
    0
    My thanks to all who pointed me in the right direction, and provided help.
    The Astribank now fails over together with DRBD and Heartbeat. The failback is done by DRBD and Heartbeat, but apparently the Astribank needs an event to do so. (Volt drop on USB port and/or disconnection to Dahdi - Which does not happen at failback - therefore this part is still manual, but I can live with that).

    I now have dual Dell severs running Elastix 1.6 with fail-over and fail-back in place. I use the Astribank with 4X PRI ports, and A2Billing for accounting - All running Sweetly!

    Thx again.
    Jan Koster
     
  9. ramoncio

    Joined:
    May 12, 2010
    Messages:
    1,663
    Likes Received:
    0
    Hi JanKoster,
    I'm glad you finally got it working.
    Twinstar is a quite recent approach to high availability from Xorcom, and it is natural if you don't get much support here. But Xorcom support is really good, I got help right from Tzafir Cohen, before he became a member of Elastix's community. I was impressed, as I had read many documents written by him.
     
  10. ashraf.khalifah@pex.ps

    Joined:
    Nov 23, 2010
    Messages:
    2
    Likes Received:
    0
    hi yaba,

    we have two xorcom VOIP boxes with centos and elastix asterisk and we want to replicate the primary box data to the secondary box to be used in case of the main box failed (and I don't want to make automatic recovery because only the primary box connected to the phone lines PRI so I want to manually recover if the primary box failed) , I read about drbd and I think it is a good choice ( do you know if there are other tools?) but we have one network card on each box so do we have to put another network card on each box? and as I told you I want to configure drbd to do one thing which is replicating the first box data to the second box.

    Thanks in advance
     
  11. JanKoster

    Joined:
    Oct 18, 2009
    Messages:
    29
    Likes Received:
    0
    Right. DRBD - you'll need two NIC's in each machine. One set for the "heartbeat" and the other for normal comms. However, the purpose of DRBD is 'auto' fail-over?
    I can not comment on any manual recovery between two machines - .... The only way I know of coming "near" this requirement you voiced, is backup and restore....
     
  12. gblanco

    Joined:
    Aug 29, 2010
    Messages:
    10
    Likes Received:
    0
    CHeck elastix cluster in wiki section
     

Share This Page