Jump to content

mattlandis

Members
  • Posts

    1,254
  • Joined

  • Last visited

Posts posted by mattlandis

  1. Hello,

     

    >Also, can you check the process list on that computer.

    This is a cleanly installed machine. When I check processes nothing is running.

     

    >I assume it does not make a difference if the call is internal, to the mailbox or external?

    I can call 1 internal extension to another on the same switch so the problem is internal.

     

    I can wireshark.

     

    matt

     

    Ok, I finally got to the bottom of this jitter.

    We were having a small 10-20% cpu spike about every 15 seconds.

    I stopped pbxnsip service and this spike continued.

    I did a little research on the service causing the spike (DPC) & found USB devices can cause this. So I unplugged the usb device in the machine and the spike stopped...along with the jitter.

     

    I then upgraded the usb driver for the device and that stopped the spiking. so all is ok now.

     

    thanks for the help,

    matt

  2. If you can easily resolve this, get a Wireshark trace and see if there is anything on the network. Maybe a packet storm every 15 seconds.

     

    I assume it does not make a difference if the call is internal, to the mailbox or external? if the problem also exist for a IVR (mailbox) call we can exclude the jitter is coming from an outside source.

     

    Also, can you check the process list on that computer. Is there anything that has a priority above normal? Only processes above normal can give you that problem.

     

    Hello,

     

    >Also, can you check the process list on that computer.

    This is a cleanly installed machine. When I check processes nothing is running.

     

    >I assume it does not make a difference if the call is internal, to the mailbox or external?

    I can call 1 internal extension to another on the same switch so the problem is internal.

     

    I can wireshark.

     

    matt

  3. Hi - Sometime\'s pictures say more than words! :D Wow! I am bowled over. You too ? Please poll!

     

    Snom_870_screen.png

     

    Snom_870_Full.png

     

    It will also be available with logo_snom_OCS_edition_blue_180px_Kopie.jpg for 2007 R1 and R2 version of

     

    mcos_logo.png

     

    Best regards,

     

    Jan

     

    Jan,

     

    It is something no one else has, and will get attention, but do people want touch screen? i'm not sure...

    I had a htc touch pda phone and went back to button phone.

     

    I saw grandstream has a phone that a mouse plugs into...actually I almost think that might be a good idea too.

     

    thanks for the nice pictures...i already posted a link to them in my blog.

    matt

  4. I discovered this too on the CS-410 with the 3144 version.

     

     

    Dial *90# and it will ask you to "Please enter the extension number"

     

    Then it will work like an intercom.

     

     

    Why has this been changed?

     

    For us when I dial *90XXX (xxx=ext.) is just rings the phone. (or do it the way you explianed)

    This WORKED for us in the previous version. (we are running windows version)

     

    we are using provisioned snom360s.

     

    tx

    matt

  5. thanks

     

    Hi,

     

    Ok, I restarted pbxnsip service and it WAS able to set the affinity according to the logfile:

     

    [5] 2009/02/21 16:55:31: Set processor affinity to 2

     

    BUT the jitter remains. It seems approx every 15 sec (very approx) it just cuts out.

    -The system has NO load when it happens

    -I've tried it with a very simple config, 2 phones, in same switch, no load, still jitter.

    -i called sales queue/73 and listened to the moh and roughly every 10-15 sec. it stutters/cuts out for a split second.

    -i've tried switching affinity to processor 1 with similar results

     

    I'd really like to resolve this,

    tx

    matt

  6. Its a little hard to catch the jitter, but i was thinking about 15sec.

     

    One problem i have in windows is I cant right click on the process and see what processor its using--i get not appropriate permissions message--soo, maybe that is an issue...i'll check.

     

    thanks

    matt

     

    also, what will trigger the "could not set affinity" log record? i just turned on logging.

     

    matt

  7. How long is the delay between the RTP packets? What is the exact time between the problems, 15 seconds?

     

    I assume the process is running with sufficient permissions and there is nothing in the log like "cound not set affinity mask".

     

    Its a little hard to catch the jitter, but i was thinking about 15sec.

     

    One problem i have in windows is I cant right click on the process and see what processor its using--i get not appropriate permissions message--soo, maybe that is an issue...i'll check.

     

    thanks

    matt

  8. Hi,

     

    We moved from a Atom1.8g/512mb ram, xp home laptop to a p4 3.4ghz/1gb ram desktop xp pro machine (clean install/nothing else on). Both had 1gb nics. On the new server we now have jitter about about every 1/4 minute. Almost undetectable. (but on vm when people let phone numbers, 1 digit is missing at times)

     

    We have set the affinity to processor 2, but the problem persists.

     

    Any other suggestions?

    tx

    matt

  9. OK I will try out the different domain for a "second" business at the same location.

     

    Can the "Seek" option be made selectable?

     

    Hi,

     

    This thread is interesting, because the last system we worked with people had a problem THAT it always started at the beginning! ;-)

     

    Maybe a setting would be good. I think "seeking into" is nice.

    It REALLY would be nice to have the "queue style" marketing message mixing available in regular music on hold.

     

    everybody wants some odd feature. ;-)

     

    tx

    matt

  10. http://forum.pbxnsip.com/index.php?showtopic=665

     

    been asking for this for over 1 year. Every small office PBX for the last 20+ years had a NIGHT button GREEN/RED for this fuction.

     

    I see several posts with dual service flags manual preceeding the auto....

     

    Everythings a work-around....

     

    Keep asking and perhaps V4.0 might have a direct solution to a old issue...

     

    Cheers,

     

    Hi,

     

    Seems like everything is in place to make this work.

    But is there some caveat that is hard to get around at the developement level?

     

    It seems dialing the service flag should flip the flag. If midnight or on/off time happens, that should put the flag to what it should be again.

     

    Isn't that easy? ;-)

    Once again, maybe there is something at developement level that makes it not that easy.

     

    tx

    matt

  11. I've seen many posts with similar symptoms reported on CS410's and we have deployed more than a few over the last year or so and have never experienced any of these issues. Lucky perhaps, but here is how we have avoided these troubles.

     

    1. We exclusively use SMART MANAGED SWITCHES and we take full advantage of all 802.1PQ tools and weighted Queues, and SNMP remote management.

    2. We NEVER.. (Seldom) .....use a NATTED FIREWALL ahead of a CS410 or any PBX. When forced to do so for one REASON only.. The client has a SINGLE IP address. We specifically use an Intertex IX series router with SIP detection. We are presently reviewing an alternative for 1 reason only. Intertex does not support SNMP access or TRAPS.

    3. ROUTERS FIREWALL ETC that have PNP autoconfigs enabled like residential LINKSYS, DLINK ETC gave a previous poster 3 weeks of really crappy performance.

    4. DON'T Log anything unless it's in the act of diagnosing or Tweaking a feature.

    5. USE SNMP wherever possible to gauge performance.

    6. USE PBXnSIP PNP with all deployments

    7. Know (Don't Assume) anything about the configuration of the analog lines by the telco provider and have the tools to verify the POTS lines are you hope they are. We see more and more POTS coming from IAD's and the config options for disconnect detections, ring volts and tones are many. REMEMBER the CS410 has a seperate 4 PORT ANALOG gateway in much the same way you would use an audiocode mp114 or other gateway.

     

     

    Good tips.

     

    We are getting a replacement cs410 and I'll report if it is ok.

     

    tx

    matt

  12. No, the blue line is the number of call legs. Usually one call equals two call legs. The red line is the number of calls, and in that graph the maximum was three calls.

     

    hi,

     

    I need to say I REALLY like this graph.

     

    A way to understand call legs is to watch the calls line on the status screen:

    Calls: 53/3 (CDR: 59) 2/4 Calls

     

    then click on the "calls" page to line up.

     

    The nice thing is we can see it.

     

    matt

  13. I read the spec and there is a lot to it. As Andrew states there will probably be some engineering time involved in this and would need to get a budget. Adding a predictive dialer is a nice feature to have but it is not a simple task.

     

    Once again, check out:

    http://www.nch.com.au/ivm/index.html

     

    It will call out, and wait to start playing your message until it hears a voice on the other end.

    Pretty sure you can do the "press 1 to transfer" also, but check this part.

     

    Very affordable. easy to setup.

    I don't sell it. ;-)

     

    tx

    matt

  14. I think 7.1.35 gve us troubles with 1-way audio. We experienced 1 way audio, and a quick hold and unhold fixed this 1 way audio problem. We downgraded 1 release. This was late last summer or early fall. 7.3.14 is now on all phones.

     

    Thanks for that input.

     

    if you think what that audiocodes parameter is let me know...that is likely the problem...because only calls thru audiocodes does it...

     

    (this problem doesn't cause us a lot of pain...but a client would not put up with it.)

     

    tx

    matt

×
×
  • Create New...