[cisco-voip] Question on CTIManager CPU % usage

[Erick Bergquist]

I can check, they had rebooted the cluster for another issue and after
2 days of being up, CTI is at 0% on both servers. I'm going to write
down the values and keep an eye on it every day or so. But it is on
5.1.1 and so probably needs 5.1.3 but I need to show them bug id's to
back it up.

On Tue, Apr 15, 2008 at 3:03 PM, Wes Sisk <wsisk at cisco.com> wrote:
> Yeah, nothing obvious in the SDL traces.  So it appears we are not in a
> loop.  The next option is a very long linked list (internal memory leak).
> There are a few ways to identify this:
>
>  1) what is the memory use now vs. when process first started up?
>  2) what is the process handle count now vs. when the process first started
> up?
>
>
>
>  /Wes
>
>  Erick Bergquist wrote:
>
> > The CTI traces were at detailed, but I'll set the SDL options and also
> > get CM Detailed and CM SDLs also and look at those and compare for
> > time period.
> >
> > Should the process id in the traces be same as process id reported in
> > RTMT PID column?  I believe RTMT reported around 40 threads for the
> > CTIManager process but I would have to check when I look at this
> > later. It's not causing a problem, that we know of, but I'm just
> > interested and want to look at the traces, etc and see if we can find
> > out why it is so high.
> >
> > Attached are the CTI Detailed and CTI SDL's I grabbed early this
> > morning; they do not have the extra flags you have mentioned below
> > set.
> >
> > Thanks
> >
> >
> > On Sat, Apr 12, 2008 at 8:39 AM, Wes Sisk <wsisk at cisco.com> wrote:
> >
> >
> > > Erick,
> > >
> > >  Good data point. That leaves 2 avenues for input to CTIManager:
> > >  1. events to CTIManager that do not show up in default trace levels
> > >  Can you try CTI traces at detailed for more information?
> > >  Make sure the SDL (SDLxxx_200_xxxxxx files) trace settings are set
> > > appropriately for CTIManager. Check/set the following "CTI Manager"
> service
> > > parameters from the CM administration. Under Service->Service Parameters
> > > select the IP address of the server. Then select the Cisco CTIManager
> > > service from the list of services. Now, click on "Advanced" button and
> > > scroll down to the group "SDL Trace Parameters". Update the following
> values
> > > :
> > >  - SdlTraceDataFlags, set to 0x00000110
> > >  - SdlTraceFlag, set to True
> > >  - SdlTraceTypeFlags, change the default setting of 0x00D0CB15 to
> 0x00F0CB15
> > >  Important: Don't forget to click the "update" button" after making
> these
> > > changes.
> > >
> > >
> > >  2. SDL signals from CallManager to CTIManager. In the CM SDL traces
> look
> > > for anything like this this regular expression:
> > >  \d,200,\d
> > >  PIDs in SDL are in the form (nodeid, processid, thread type, thread
> > > instance)
> > >  so (1,100,*,*) are all threads in ccm.exe on node 1.
> > >  (1,200,*,*) are all threads in ctimanager on node 1.
> > >
> > >  If CM is sending or receiving excess signals with CTIManager this will
> tell
> > > why kind of signals and from that we can figure out the activity.
> > >
> > >  /Wes
> > >
> > >
> > >
> > >  Erick Bergquist wrote:
> > >
> > >
> > >
> > > > The CTI SDI and SDL traces are just polls and keepalive traffic for
> > > > most part. It is early am hours and CTI Manager is taking 45-55% of
> > > > CPU and there are 0 active calls cluster wide.  I also disabled CTI
> > > > traces and same %. There are 135 Devices open for CTI in performance
> > > > monitor. Could it be Mobility Manager and IPCC and CER monitoring the
> > > > lines? Theres 13 IPCC agents and around 80 mobility devices, rest is
> > > > AC and CER, some Cisco TSP users.
> > > >
> > > > Thanks, Erick
> > > >
> > > > On Thu, Apr 10, 2008 at 12:40 PM, Erick Bergquist <erickbee at gmail.com>
> > > >
> > > >
> > > wrote:
> > >
> > >
> > > >
> > > >
> > > > > I am digging through some of those now. It is a MCS-7815I so has one
> > > > >  processor I believe.
> > > > >
> > > > >
> > > > >
> > > > >  On Thu, Apr 10, 2008 at 12:32 PM, Wes Sisk <wsisk at cisco.com> wrote:
> > > > >  > what platform? If multiple CPU or multi core I suspect you're
> > > > >
> > > > >
> > > >
> > > actually
> > >
> > >
> > > >
> > > > >  > seeing CTIManager use 100% of one core/processor.
> > > > >  >
> > > > >  >  CTI Manager SDI and CTIManager SDL traces will show why.
> > > > >  >
> > > > >  >  /Wes
> > > > >  >
> > > > >  >  Erick Bergquist wrote:
> > > > >  >
> > > > >  > >
> > > > >  > >
> > > > >  > >
> > > > >  > > What can I look at to see what is causing the CPU % to be
> around
> > > > >
> > > > >
> > > >
> > > 50%
> > >
> > >
> > > >
> > > > >  > > for this one process (RTMT), or if it is normal for the setup?
> It
> > > > >
> > > > >
> > > >
> > > is
> > >
> > >
> > > >
> > > > >  > > version 5.1.1 and using IPCC, CER, Mobility Manager, Attendant
> > > > >  > > Console, and a few users have CiscoTSP on their desktop. There
> are
> > > > >
> > > > >
> > > >
> > > 130
> > >
> > >
> > > >
> > > > >  > > lines controlled by CTI and about 3 dozen devices. There are
> only
> > > > >
> > > > >
> > > >
> > > 10
> > >
> > >
> > > >
> > > > >  > > calls active cluster wide.
> > > > >  > > _______________________________________________
> > > > >  > > cisco-voip mailing list
> > > > >  > > cisco-voip at puck.nether.net
> > > > >  > > https://puck.nether.net/mailman/listinfo/cisco-voip
> > > > >  > >
> > > > >  > >
> > > > >  >
> > > > >
> > > > >
> > > > >
> > > > >
> > > > >
> > > >
> > >
> >
>