asterisk users - Nov 2006 - Stable clock with 2.6 and without Digium hardware.

Anybody sucessfully got stable 1000Hz clock without Digium harware and kernel
2.6?
We need to consult some peoples how to clock asterisk stable with exactly 1000
Hz without much kernel/drives patching/tweaking.

Some test results we made so far:

2.6 with digium card - stable 1000 Hz.
2.6 with ztdummy - uses RTC and the clock is 1024, not 1000.
2.6 with some Realtime kernel patch - provides stable 1000 Hz for some time, but
in moments stops/misses interrupts/goes away from 1000 Hz
2.6 with ztdummy USB_UHCI - don't works, needs some tweaking. Somebody knows
good patch for it?
2.6 with ztdynamic as primary clock sources - some issues with 2.6 (ztdynamic
not ported well to 2.6?) with the mainstream versions, somehow patches
solves it.
2.6 with kernel clock - needs kernel recompiling and work stable with switched
off kernel Preemption. Long time tests in progress now.

2.4 with digium card - stable 1000 Hz
2.4 with ztdummy UHCI - stable 1000 Hz
2.4 with ztdynamic clock source - stable 1000 Hz/Depends on network conditions.

Can you tell us how you do the testing ?

Zoa.


Anton Tinchev wrote:
> Anybody sucessfully got stable 1000Hz clock without Digium harware and 
> kernel 2.6?
> We need to consult some peoples how to clock asterisk stable with 
> exactly 1000 Hz without much kernel/drives patching/tweaking.
>
> Some test results we made so far:
>
> 2.6 with digium card - stable 1000 Hz.
> 2.6 with ztdummy - uses RTC and the clock is 1024, not 1000.
> 2.6 with some Realtime kernel patch - provides stable 1000 Hz for some 
> time, but in moments stops/misses interrupts/goes away from 1000 Hz
> 2.6 with ztdummy USB_UHCI - don't works, needs some tweaking. Somebody 
> knows good patch for it?
> 2.6 with ztdynamic as primary clock sources - some issues with 2.6 
> (ztdynamic not ported well to 2.6?) with the mainstream versions, 
> somehow patches solves it.
> 2.6 with kernel clock - needs kernel recompiling and work stable with 
> switched off kernel Preemption. Long time tests in progress now.
>
> 2.4 with digium card - stable 1000 Hz
> 2.4 with ztdummy UHCI - stable 1000 Hz
> 2.4 with ztdynamic clock source - stable 1000 Hz/Depends on network 
> conditions.
>
>
> _______________________________________________
> --Bandwidth and Colocation provided by Easynews.com --
>
> asterisk-users mailing list
> To UNSUBSCRIBE or update options visit:
>   http://lists.digium.com/mailman/listinfo/asterisk-users

Zoa wrote:
> Can you tell us how you do the testing ?
3-4 different ways. All gives same results, so test are pretty valid.

1. Interrupt counting inside the PC.
2. TDMoE packet counting on the switch.
3. External TDMoE equipment connected thru extreme network swich.
The card of the PC and the device only connected to the switch. The switch
filters all packets except TDMoE to de device.
Calibrated oscilloscope conected to the interupt leg of the network chip. All
coalescing/etc disabling.
4. Diagnostic results from firmware of the device.
5. ToDo test - oscilloscope directly conected to pads inside the PC, but needs
mechanical work for each platform/type.
6. ToDo test - some driver relays the clock to simple hardware card in the PC
and oscilloscope connected to it.

All 4 tests reports same clock difference/clock misses etc.
Tested at 3-4 types of hardware - different chipsets/processors. Same results.
The are sheduled tests to around 30 more platforms, but pretty sure
that the results will be similar.

P.S.
The device is TDMoE FXS/FXO modular channel bank currently ending development
and starting production.

On Fri, Nov 10, 2006 at 03:28:09PM +0200, Anton Tinchev
wrote:
> Zoa wrote:
> 
> >Can you tell us how you do the testing ?
> 3-4 different ways. All gives same results, so test are pretty valid.
> 
> 1. Interrupt counting inside the PC.
> 2. TDMoE packet counting on the switch.
> 3. External TDMoE equipment connected thru extreme network swich.
> The card of the PC and the device only connected to the switch. The switch 
> filters all packets except TDMoE to de device.
> Calibrated oscilloscope conected to the interupt leg of the network chip. 
> All coalescing/etc disabling.
> 4. Diagnostic results from firmware of the device.
> 5. ToDo test - oscilloscope directly conected to pads inside the PC, but 
> needs mechanical work for each platform/type.
> 6. ToDo test - some driver relays the clock to simple hardware card in the 
> PC and oscilloscope connected to it.
How about meassuring it directly? For starters, take a look at zttest.c .

(Though it could use some slightly better accuracy).

-- 
               Tzafrir Cohen       
icq#16849755                    jabber:tzafrir@jabber.org
+972-50-7952406           mailto:tzafrir.cohen@xorcom.com       
http://www.xorcom.com  iax:guest@local.xorcom.com/tzafrir

On Fri, Nov 10, 2006 at 03:04:46PM +0200, Anton Tinchev
wrote:
> Anybody sucessfully got stable 1000Hz clock without Digium harware and 
> kernel 2.6?
> We need to consult some peoples how to clock asterisk stable with exactly 
> 1000 Hz without much kernel/drives patching/tweaking.
> 
> Some test results we made so far:
> 
> 2.6 with digium card - stable 1000 Hz.
Which card, BTW?
> 2.6 with ztdummy - uses RTC and the clock is 1024, not 1000.
> 2.6 with some Realtime kernel patch - provides stable 1000 Hz for some 
> time, but in moments stops/misses interrupts/goes away from 1000 Hz
> 2.6 with ztdummy USB_UHCI - don't works, needs some tweaking. Somebody 
> knows good patch for it?
> 2.6 with ztdynamic as primary clock sources - some issues with 2.6 
> (ztdynamic not ported well to 2.6?) with the mainstream versions, somehow 
> patches solves it.
> 2.6 with kernel clock - needs kernel recompiling and work stable with 
> switched off kernel Preemption. Long time tests in progress now.
Why does preemption conflict with HZ=1000 ? Totally disable preemption? 

On what system(s)?
> 
> 2.4 with digium card - stable 1000 Hz
> 2.4 with ztdummy UHCI - stable 1000 Hz
> 2.4 with ztdynamic clock source - stable 1000 Hz/Depends on network 
> conditions.
-- 
               Tzafrir Cohen       
icq#16849755                    jabber:tzafrir@jabber.org
+972-50-7952406           mailto:tzafrir.cohen@xorcom.com       
http://www.xorcom.com  iax:guest@local.xorcom.com/tzafrir

<snip>
Anton wrote:
> 2.6 with kernel clock - needs kernel recompiling and
> work stable with switched off kernel Preemption. Long
> time tests in progress now.
How are you doing this?  I saw a couple developers talk
about rewriting ztdummy yo use the new hi-res kernel timers
(kernel>2.6.17), but did not notice any patches as a result.

I've been pretty happy with the RTC-based ztdummy, except
that the RTC can be a bit of a pig when run at 8192 (~8% cpu
load).  Hi-res timers firing every 1ms should produce a
lighter load and be more accurate....

Dan

> Which card, BTW?
TDM400 analog

HI, all 
Can I disable send e-mail feature in the voicemail application?

just dont enter any email address while creating extension / mailbox ;)

On 14/11/06, Ma Zhiyong <mazhiyong@sohu.com>
wrote:
>
> HI, all
> Can I disable send e-mail feature in the voicemail application?
> _______________________________________________
> --Bandwidth and Colocation provided by Easynews.com --
>
> asterisk-users mailing list
> To UNSUBSCRIBE or update options visit:
>    http://lists.digium.com/mailman/listinfo/asterisk-users
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
http://lists.digium.com/pipermail/asterisk-users/attachments/20061114/f780077d/attachment.htm

Yes. Just dont add a line in voicemail.conf with an email address.

----- Original Message ----- 
From: "Ma Zhiyong" <mazhiyong@sohu.com>
To: "Asterisk Users Mailing List - Non-Commercial Discussion" 
<asterisk-users@lists.digium.com>
Sent: Tuesday, November 14, 2006 1:34 PM
Subject: [asterisk-users] Can I disable send e-mail feature in the 
voicemailapplication?

> HI, all
> Can I disable send e-mail feature in the voicemail application?

--------------------------------------------------------------------------------

> _______________________________________________
> --Bandwidth and Colocation provided by Easynews.com --
>
> asterisk-users mailing list
> To UNSUBSCRIBE or update options visit:
>   http://lists.digium.com/mailman/listinfo/asterisk-users
>