Re: [asterisk-users] Maximum number of PRI calls on 1 asterisk box (no HW echo)
On 25 March 2010 02:42, James Lamanna wrote: > Hi, > I have a TE410P T1 (1st gen) card and I'm seeing interesting errors of > D-Channels going down and then coming back up (See below). > > Read all the discussion about many spans - and I've run 16 E1 spans in one box, and run 8 spans under 200+ concurrent calls. Your 1st gen TE410 card is very old and I'd suggest to contact Digium about a firmware upgrade or a hardware upgrade. As for the spans going down:- 1) Make sure you are syncing your clock to your telco (span=1,1,0,...) 2) Make sure you are using the right IDE driver module for your chipset and not the generic one 3) Avoid long runs (>25m?) of unscreened cable on the T1/E1 span Regards, Steve -- _ -- Bandwidth and Colocation Provided by http://www.api-digital.com -- New to Asterisk? Join us for a live introductory webinar every Thurs: http://www.asterisk.org/hello asterisk-users mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users
Re: [asterisk-users] Maximum number of PRI calls on 1 asterisk box (no HW echo)
James Lamanna wrote: > Hi, > Does anyone have any good empirical data suggesting what the maximum > number of PRI calls (incoming and outgoing) > without hardware echo cancellation can be handled on a single box is? > I have a TE410P T1 (1st gen) card and I'm seeing interesting errors of > D-Channels going down and then coming back up (See below). > > I've looked at the number of simultaneous calls at each of these > points, and each time the span seems to > have around 21-23 calls, and the total number of calls ranges between 47 and > 53. > I'm trying to figure out if this is a load issue or an issue on the > provider side, though my provider says they > do not see any errors on any of the T1s. > Could this be some sort of hardware interrupt problem? If so, how can I check? > > The specs of the machine are, Dual Xeon 2.80Ghz (both single core but w/HT) > 4GB memory. > Running asterisk 1.4.26.3 (32-bit) > with libpri-1.4.7 and zaptel-1.4.12.9 > > Thanks. > > -- James > > Please CC me on responses. > > > [Mar 22 09:45:00] VERBOSE[8887] logger.c: == Primary D-Channel on span 2 > down > [Mar 22 09:45:00] WARNING[8887] chan_dahdi.c: No D-channels available! > Using Primary channel 48 as D-channel anyway! > [Mar 22 09:45:00] VERBOSE[8887] logger.c: == Primary D-Channel on span 2 up > [Mar 22 09:59:23] VERBOSE[8886] logger.c: == Primary D-Channel on span 1 > down > [Mar 22 09:59:23] WARNING[8886] chan_dahdi.c: No D-channels available! > Using Primary channel 24 as D-channel anyway! > [Mar 22 09:59:23] VERBOSE[8886] logger.c: == Primary D-Channel on span 1 up > [Mar 22 09:59:23] VERBOSE[8886] logger.c: == Primary D-Channel on span 1 > down > [Mar 22 09:59:23] WARNING[8886] chan_dahdi.c: No D-channels available! > Using Primary channel 24 as D-channel anyway! > [Mar 22 09:59:23] VERBOSE[8886] logger.c: == Primary D-Channel on span 1 up > [Mar 22 10:36:11] VERBOSE[8886] logger.c: == Primary D-Channel on span 1 > down > [Mar 22 10:36:11] WARNING[8886] chan_dahdi.c: No D-channels available! > Using Primary channel 24 as D-channel anyway! > [Mar 22 10:36:11] VERBOSE[8886] logger.c: == Primary D-Channel on span 1 up > [Mar 22 10:36:11] VERBOSE[8886] logger.c: == Primary D-Channel on span 1 > down > [Mar 22 10:36:11] WARNING[8886] chan_dahdi.c: No D-channels available! > Using Primary channel 24 as D-channel anyway! > [Mar 22 10:36:11] VERBOSE[8886] logger.c: == Primary D-Channel on span 1 up > [Mar 22 10:44:36] NOTICE[] chan_dahdi.c: PRI got event: HDLC Bad > FCS (8) on Primary D-channel of span 3 > [Mar 22 10:45:44] NOTICE[8886] chan_dahdi.c: PRI got event: HDLC Bad > FCS (8) on Primary D-channel of span 1 > [Mar 22 10:59:33] NOTICE[8887] chan_dahdi.c: PRI got event: HDLC Abort > (6) on Primary D-channel of span 2 > [Mar 22 11:30:53] VERBOSE[8886] logger.c: == Primary D-Channel on span 1 > down > [Mar 22 11:30:53] WARNING[8886] chan_dahdi.c: No D-channels available! > Using Primary channel 24 as D-channel anyway! > [Mar 22 11:30:53] VERBOSE[8886] logger.c: == Primary D-Channel on span 1 up > [Mar 22 15:34:28] VERBOSE[8887] logger.c: == Primary D-Channel on span 2 > down > [Mar 22 15:34:28] WARNING[8887] chan_dahdi.c: No D-channels available! > Using Primary channel 48 as D-channel anyway! > [Mar 22 15:34:28] VERBOSE[8887] logger.c: == Primary D-Channel on span 2 up > > I believe your problem is IRQ sharing which is why you are getting the "...HDLC Bad FCS..." notices in your debug. Make sure the card is not sharing IRQs with the network card(s), or the hard drive(s). You MIGHT get by with it sharing with other devices. Upgrade to a newer Digium card that is not as picky with IRQ sharing or switch brands. I've had several TE410 cards and they don't like to share. Alex -- _ -- Bandwidth and Colocation Provided by http://www.api-digital.com -- New to Asterisk? Join us for a live introductory webinar every Thurs: http://www.asterisk.org/hello asterisk-users mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users
Re: [asterisk-users] Maximum number of PRI calls on 1 asterisk box (no HW echo)
2010/3/25 Zeeshan Zakaria : > Tzafrir, so you have actually worked with more than 192 concurrent zap > channels, which means more than 8 spans, on a single server, and can verify > that it actually works without freezing asterisk. As I have written before - I did use 8 E1 in one machine quite often. So I can confirm that it runs with at least 240 B-channels. If my mind serves me right we even had one installation with 480 channels - not with TE410P though. Chris -- _ -- Bandwidth and Colocation Provided by http://www.api-digital.com -- New to Asterisk? Join us for a live introductory webinar every Thurs: http://www.asterisk.org/hello asterisk-users mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users
Re: [asterisk-users] Maximum number of PRI calls on 1 asterisk box (no HW echo)
2010/3/25 Steve Edwards : > On Thu, 25 Mar 2010, Tzafrir Cohen wrote: > > [snipping a lot of interesting technical and historical details] > >> As you can see, there's actually a limit at the DAHDI level. >> DAHDI_MAX_SPANS, which is 128. Likewise there's DAHDI_MAX_CHANS which is >> 1024. That's as many channels that you can have. > > All good stuff. I would just question the sanity of putting 1024 eggs in a > single basket. Well - think of having 2048 eggs and only two hands. ;-) Chris -- _ -- Bandwidth and Colocation Provided by http://www.api-digital.com -- New to Asterisk? Join us for a live introductory webinar every Thurs: http://www.asterisk.org/hello asterisk-users mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users
Re: [asterisk-users] Maximum number of PRI calls on 1 asterisk box (no HW echo)
Tzafrir, so you have actually worked with more than 192 concurrent zap channels, which means more than 8 spans, on a single server, and can verify that it actually works without freezing asterisk. I really need specs for this system. I'll recompile zaptel, no problem, but it'll save me one extra server, for a customer for whom I am doing the implementation these days. I'll get the new cards as well, though Rhino cards are very good cards, but I need more spans per server, which will actually work in production not freezing asterisk. -- Zeeshan A Zakaria On 2010-03-25 1:45 PM, "Tzafrir Cohen" wrote: On Thu, Mar 25, 2010 at 10:24:41AM -0600, Carlos Chavez wrote: > On Thu, 2010-03-25 at 17:11 +0100, ... The "limit" of 250 (actually 249. It was 250 originally and changed to 249 when /dev/zap/transcode was introduced) zap channels is only that of the device files that are generated. But userspace programs normally use /dev/dahdi/channel to ask for a channel (using the ioctl DAHDI_SPECIFY) and don't need to open /dev/dahdi/NNN directly (but if that really bothers you, that limitation has been removed in the branhc /team/tzafrir/sysfs ). Asterisk, though, has an arrach of 32 'spans' (PRI spans). This means that if you try using a PRI span number whose number >= 32 you'll get an annoying error. Someone who encontered that error early on was KP Junghanns. He added support for BRI devices to chan_zap. Thus he had much smaller spans. And could afford inclding many of them (a single octobri card has 8 spans). Which is why the bristuff patch includes the following: -#define NUM_SPANS 32 +#define NUM_SPANS 128 /*!<"32 spans", muahahaha, us alaws like to have some more... */ In the Debian package I incllude a slightly different variant of it: -#define NUM_SPANS 32 +#define NUM_SPANS DAHDI_MAX_SPANS As you can see, there's actually a limit at the DAHDI level. DAHDI_MAX_SPANS, which is 128. Likewise there's DAHDI_MAX_CHANS which is 1024. That's as many channels that you can have. (Disclaimer: I may be off by one here and there, as DAHDI span numbers and channel numbers are 1-based rather than 0-based) If you need more, you'll just have to rebuild DAHDI (been there, done that ;-) -- Tzafrir Cohen icq#16849755 jabber:tzafrir.co...@xorcom.com +972-50-7952406 mailto:tzafrir.co...@xorcom.com http://www.xorcom.com iax:gu...@local.xorcom.com/tzafrir -- _ -- Bandwidth and Colocation Pr... -- _ -- Bandwidth and Colocation Provided by http://www.api-digital.com -- New to Asterisk? Join us for a live introductory webinar every Thurs: http://www.asterisk.org/hello asterisk-users mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users
Re: [asterisk-users] Maximum number of PRI calls on 1 asterisk box (no HW echo)
On Thu, 25 Mar 2010, Tzafrir Cohen wrote: [snipping a lot of interesting technical and historical details] > As you can see, there's actually a limit at the DAHDI level. > DAHDI_MAX_SPANS, which is 128. Likewise there's DAHDI_MAX_CHANS which is > 1024. That's as many channels that you can have. All good stuff. I would just question the sanity of putting 1024 eggs in a single basket. -- Thanks in advance, - Steve Edwards sedwa...@sedwards.com Voice: +1-760-468-3867 PST Newline Fax: +1-760-731-3000 -- _ -- Bandwidth and Colocation Provided by http://www.api-digital.com -- New to Asterisk? Join us for a live introductory webinar every Thurs: http://www.asterisk.org/hello asterisk-users mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users
Re: [asterisk-users] Maximum number of PRI calls on 1 asterisk box (no HW echo)
On Thu, Mar 25, 2010 at 10:24:41AM -0600, Carlos Chavez wrote: > On Thu, 2010-03-25 at 17:11 +0100, Christian Victor wrote: > > Hi James, > > > > we did sucessfully run two TE410P with 8xE1 in a decent Intel P4-3200 > > machine with quite heavy line usage. No codec conversion course. > > > > I don't believe that there is a hard limit of E1s coded into Asterisk. > > But the maximum lines you can squeeze out of your specific hardware > > depends on so many factors (asterisk version, hardware etc). > > > > Chris > > > If you ever take the DCAP training they use a single server with 16 T1 > Spans (4 quad cards) and a 24 port analog card to provide connectivity > to the participants. That means that they have 5 cards and 17 spans for > all ports. Obviously there is not much load on the system because they > are not using all channels at once. Asterisk can handle all those ports > so there does not seem to be a set limit as there was in the 1.0 days > where you could not have more 250 zap devices. The "limit" of 250 (actually 249. It was 250 originally and changed to 249 when /dev/zap/transcode was introduced) zap channels is only that of the device files that are generated. But userspace programs normally use /dev/dahdi/channel to ask for a channel (using the ioctl DAHDI_SPECIFY) and don't need to open /dev/dahdi/NNN directly (but if that really bothers you, that limitation has been removed in the branhc /team/tzafrir/sysfs ). Asterisk, though, has an arrach of 32 'spans' (PRI spans). This means that if you try using a PRI span number whose number >= 32 you'll get an annoying error. Someone who encontered that error early on was KP Junghanns. He added support for BRI devices to chan_zap. Thus he had much smaller spans. And could afford inclding many of them (a single octobri card has 8 spans). Which is why the bristuff patch includes the following: -#define NUM_SPANS 32 +#define NUM_SPANS 128 /*!<"32 spans", muahahaha, us alaws like to have some more... */ In the Debian package I incllude a slightly different variant of it: -#define NUM_SPANS 32 +#define NUM_SPANS DAHDI_MAX_SPANS As you can see, there's actually a limit at the DAHDI level. DAHDI_MAX_SPANS, which is 128. Likewise there's DAHDI_MAX_CHANS which is 1024. That's as many channels that you can have. (Disclaimer: I may be off by one here and there, as DAHDI span numbers and channel numbers are 1-based rather than 0-based) If you need more, you'll just have to rebuild DAHDI (been there, done that ;-) -- Tzafrir Cohen icq#16849755 jabber:tzafrir.co...@xorcom.com +972-50-7952406 mailto:tzafrir.co...@xorcom.com http://www.xorcom.com iax:gu...@local.xorcom.com/tzafrir -- _ -- Bandwidth and Colocation Provided by http://www.api-digital.com -- New to Asterisk? Join us for a live introductory webinar every Thurs: http://www.asterisk.org/hello asterisk-users mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users
Re: [asterisk-users] Maximum number of PRI calls on 1 asterisk box (no HW echo)
Does anybody know specs of their server, cards, and versions of asterisk, zaptel and libpri? On 2010-03-25 12:28 PM, "Carlos Chavez" wrote: On Thu, 2010-03-25 at 17:11 +0100, Christian Victor wrote: > Hi James, > > we did sucessfully run t... If you ever take the DCAP training they use a single server with 16 T1 Spans (4 quad cards) and a 24 port analog card to provide connectivity to the participants. That means that they have 5 cards and 17 spans for all ports. Obviously there is not much load on the system because they are not using all channels at once. Asterisk can handle all those ports so there does not seem to be a set limit as there was in the 1.0 days where you could not have more 250 zap devices. -- Telecomunicaciones Abiertas de México S.A. de C.V. Carlos Chávez Prats Director de Tecnología +52-55-91169161 ext 2001 -- _ -- Bandwidth and Colocation Provided by http://www.api-digital.com -- New to Asterisk? Join us for a live introductory webinar every Thurs: http://www.asterisk.org/hello asterisk-users mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users -- _ -- Bandwidth and Colocation Provided by http://www.api-digital.com -- New to Asterisk? Join us for a live introductory webinar every Thurs: http://www.asterisk.org/hello asterisk-users mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users
Re: [asterisk-users] Maximum number of PRI calls on 1 asterisk box (no HW echo)
On Thu, 2010-03-25 at 17:11 +0100, Christian Victor wrote: > Hi James, > > we did sucessfully run two TE410P with 8xE1 in a decent Intel P4-3200 > machine with quite heavy line usage. No codec conversion course. > > I don't believe that there is a hard limit of E1s coded into Asterisk. > But the maximum lines you can squeeze out of your specific hardware > depends on so many factors (asterisk version, hardware etc). > > Chris > If you ever take the DCAP training they use a single server with 16 T1 Spans (4 quad cards) and a 24 port analog card to provide connectivity to the participants. That means that they have 5 cards and 17 spans for all ports. Obviously there is not much load on the system because they are not using all channels at once. Asterisk can handle all those ports so there does not seem to be a set limit as there was in the 1.0 days where you could not have more 250 zap devices. -- Telecomunicaciones Abiertas de México S.A. de C.V. Carlos Chávez Prats Director de Tecnología +52-55-91169161 ext 2001 signature.asc Description: This is a digitally signed message part -- _ -- Bandwidth and Colocation Provided by http://www.api-digital.com -- New to Asterisk? Join us for a live introductory webinar every Thurs: http://www.asterisk.org/hello asterisk-users mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users
Re: [asterisk-users] Maximum number of PRI calls on 1 asterisk box (no HW echo)
Chris, I am running 8 spans on a few servers. It is when you go beyond that. If anyone is running 9 or more spans successfully, please let us know their configuration, it'll be very helpful. -- Zeeshan A Zakaria On 2010-03-25 12:19 PM, "Christian Victor" wrote: Hi James, we did sucessfully run two TE410P with 8xE1 in a decent Intel P4-3200 machine with quite heavy line usage. No codec conversion course. I don't believe that there is a hard limit of E1s coded into Asterisk. But the maximum lines you can squeeze out of your specific hardware depends on so many factors (asterisk version, hardware etc). Chris 2010/3/25 James Lamanna : > Hi, > Does anyone have any good empirical data suggesting what the maximum > number of PRI calls (... -- _ -- Bandwidth and Colocation Provided by http://www.api-digital.com -- New to Asterisk? Join us for a live introductory webinar every Thurs: http://www.asterisk.org/hello asterisk-users mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users
Re: [asterisk-users] Maximum number of PRI calls on 1 asterisk box (no HW echo)
Hi James, we did sucessfully run two TE410P with 8xE1 in a decent Intel P4-3200 machine with quite heavy line usage. No codec conversion course. I don't believe that there is a hard limit of E1s coded into Asterisk. But the maximum lines you can squeeze out of your specific hardware depends on so many factors (asterisk version, hardware etc). Chris 2010/3/25 James Lamanna : > Hi, > Does anyone have any good empirical data suggesting what the maximum > number of PRI calls (incoming and outgoing) > without hardware echo cancellation can be handled on a single box is? > I have a TE410P T1 (1st gen) card and I'm seeing interesting errors of > D-Channels going down and then coming back up (See below). > > I've looked at the number of simultaneous calls at each of these > points, and each time the span seems to > have around 21-23 calls, and the total number of calls ranges between 47 and > 53. > I'm trying to figure out if this is a load issue or an issue on the > provider side, though my provider says they > do not see any errors on any of the T1s. > Could this be some sort of hardware interrupt problem? If so, how can I check? > > The specs of the machine are, Dual Xeon 2.80Ghz (both single core but w/HT) > 4GB memory. > Running asterisk 1.4.26.3 (32-bit) > with libpri-1.4.7 and zaptel-1.4.12.9 > > Thanks. > > -- James > > Please CC me on responses. > > > [Mar 22 09:45:00] VERBOSE[8887] logger.c: == Primary D-Channel on span 2 > down > [Mar 22 09:45:00] WARNING[8887] chan_dahdi.c: No D-channels available! > Using Primary channel 48 as D-channel anyway! > [Mar 22 09:45:00] VERBOSE[8887] logger.c: == Primary D-Channel on span 2 up > [Mar 22 09:59:23] VERBOSE[8886] logger.c: == Primary D-Channel on span 1 > down > [Mar 22 09:59:23] WARNING[8886] chan_dahdi.c: No D-channels available! > Using Primary channel 24 as D-channel anyway! > [Mar 22 09:59:23] VERBOSE[8886] logger.c: == Primary D-Channel on span 1 up > [Mar 22 09:59:23] VERBOSE[8886] logger.c: == Primary D-Channel on span 1 > down > [Mar 22 09:59:23] WARNING[8886] chan_dahdi.c: No D-channels available! > Using Primary channel 24 as D-channel anyway! > [Mar 22 09:59:23] VERBOSE[8886] logger.c: == Primary D-Channel on span 1 up > [Mar 22 10:36:11] VERBOSE[8886] logger.c: == Primary D-Channel on span 1 > down > [Mar 22 10:36:11] WARNING[8886] chan_dahdi.c: No D-channels available! > Using Primary channel 24 as D-channel anyway! > [Mar 22 10:36:11] VERBOSE[8886] logger.c: == Primary D-Channel on span 1 up > [Mar 22 10:36:11] VERBOSE[8886] logger.c: == Primary D-Channel on span 1 > down > [Mar 22 10:36:11] WARNING[8886] chan_dahdi.c: No D-channels available! > Using Primary channel 24 as D-channel anyway! > [Mar 22 10:36:11] VERBOSE[8886] logger.c: == Primary D-Channel on span 1 up > [Mar 22 10:44:36] NOTICE[] chan_dahdi.c: PRI got event: HDLC Bad > FCS (8) on Primary D-channel of span 3 > [Mar 22 10:45:44] NOTICE[8886] chan_dahdi.c: PRI got event: HDLC Bad > FCS (8) on Primary D-channel of span 1 > [Mar 22 10:59:33] NOTICE[8887] chan_dahdi.c: PRI got event: HDLC Abort > (6) on Primary D-channel of span 2 > [Mar 22 11:30:53] VERBOSE[8886] logger.c: == Primary D-Channel on span 1 > down > [Mar 22 11:30:53] WARNING[8886] chan_dahdi.c: No D-channels available! > Using Primary channel 24 as D-channel anyway! > [Mar 22 11:30:53] VERBOSE[8886] logger.c: == Primary D-Channel on span 1 up > [Mar 22 15:34:28] VERBOSE[8887] logger.c: == Primary D-Channel on span 2 > down > [Mar 22 15:34:28] WARNING[8887] chan_dahdi.c: No D-channels available! > Using Primary channel 48 as D-channel anyway! > [Mar 22 15:34:28] VERBOSE[8887] logger.c: == Primary D-Channel on span 2 up > > -- > _ > -- Bandwidth and Colocation Provided by http://www.api-digital.com -- > New to Asterisk? Join us for a live introductory webinar every Thurs: > http://www.asterisk.org/hello > > asterisk-users mailing list > To UNSUBSCRIBE or update options visit: > http://lists.digium.com/mailman/listinfo/asterisk-users > -- _ -- Bandwidth and Colocation Provided by http://www.api-digital.com -- New to Asterisk? Join us for a live introductory webinar every Thurs: http://www.asterisk.org/hello asterisk-users mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users
Re: [asterisk-users] Maximum number of PRI calls on 1 asterisk box (no HW echo)
Zeeshan A Zakaria wrote: >On Wed, Mar 24, 2010 at 5:42 PM, James Lamanna wrote: [snip] >> >> The specs of the machine are, Dual Xeon 2.80Ghz (both single core but w/HT) >> 4GB memory. >> Running asterisk 1.4.26.3 (32-bit) >> with libpri-1.4.7 and zaptel-1.4.12.9 > >So I think it is not your T1 card but some software/driver issue. Did you >upgrade anything recently on this server? I actually downgraded this server from 1.6.0.12 to the above configuration. I was having issues with spans locking all their channels with 1.6.0.12. -- James ** Please CC me directly on all responses - I am subscribed in digest mode ** -- _ -- Bandwidth and Colocation Provided by http://www.api-digital.com -- New to Asterisk? Join us for a live introductory webinar every Thurs: http://www.asterisk.org/hello asterisk-users mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users
Re: [asterisk-users] Maximum number of PRI calls on 1 asterisk box (no HW echo)
A good idea is to call sangoma tech support or email them with your config but as per my knowledge and experience and as per Rhino's tech support and Digium techsupport suggestions which I got when dealing with a lot of spans, you can't have 10 PRIs running under one asterisk. You'll need to physically remove one 4xPRI card, and set it up on a separate server, and configure your dialplan accordingly. On 2010-03-24 11:14 PM, "Ryan Wagoner" wrote: I'm running 2 Sangoma A104D cards and a Sangoma A102D card for a total of 10 PRIs. Right now I have 9 PRIs configured as 3 DAHDI groups. 3 PRIs are in a group for local calls, 2 are grouped for long distance, and the last 4 connect to our Toshiba PBX for users still on digital phones. These cards are in a Dell R710 with two quad core CPU's. System load hovers just below 1 with 30-40 concurrent calls. Half the calls are to SIP phones connected over OpenVPN running on the server. The other half are DAHDI to DAHDI bridged to the PBX. Software wise I am running CentOS 5.4 x86_64, Asterisk 1.6.1.18, DAHDI 2.2.1, libPRI 1.4.10.2, and Sangoma wanpipe driver 3.5.6. I had some issues with newer wanpipe drivers and kernel soft locks. I also had a PCI dma timeout issue which required a Sangoma firmware update. Since then it has been rock solid since with 22 days of uptime. Ryan On Wed, Mar 24, 2010 at 9:53 PM, Zeeshan Zakaria wrote: > In theory asterisk c... -- _ -- Bandwidth and Colocation Pr... -- _ -- Bandwidth and Colocation Provided by http://www.api-digital.com -- New to Asterisk? Join us for a live introductory webinar every Thurs: http://www.asterisk.org/hello asterisk-users mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users
Re: [asterisk-users] Maximum number of PRI calls on 1 asterisk box (no HW echo)
I'm running 2 Sangoma A104D cards and a Sangoma A102D card for a total of 10 PRIs. Right now I have 9 PRIs configured as 3 DAHDI groups. 3 PRIs are in a group for local calls, 2 are grouped for long distance, and the last 4 connect to our Toshiba PBX for users still on digital phones. These cards are in a Dell R710 with two quad core CPU's. System load hovers just below 1 with 30-40 concurrent calls. Half the calls are to SIP phones connected over OpenVPN running on the server. The other half are DAHDI to DAHDI bridged to the PBX. Software wise I am running CentOS 5.4 x86_64, Asterisk 1.6.1.18, DAHDI 2.2.1, libPRI 1.4.10.2, and Sangoma wanpipe driver 3.5.6. I had some issues with newer wanpipe drivers and kernel soft locks. I also had a PCI dma timeout issue which required a Sangoma firmware update. Since then it has been rock solid since with 22 days of uptime. Ryan On Wed, Mar 24, 2010 at 9:53 PM, Zeeshan Zakaria wrote: > In theory asterisk can handle 8 spans, which means 192 concurrent calls if > all spans are T1. In reality after 6 spans it starts giving problems. > > I recently shipped a server to a client in Taiwan, and before shipment it > was put under a load of 100 plus concurrent calls (zap channels) for two > whole weeks without any break. It had 6 spans, Rhino T1, FXO and FXS cards > all with on board EC. The server's processors combined were hardly 7% in use > and no call quality issues. For the same client another server with 9 spans > were causing asterisk to freeze once there are more than 96 calls, i.e. it > going to 1st span of the second T1 card, which made it 9th span considering > FXO and FXS spans, but server load remained low as usual. > > So I think it is not your T1 card but some software/driver issue. Did you > upgrade anything recently on this server? > > -- > Zeeshan A Zakaria > > On 2010-03-24 8:49 PM, "James Lamanna" wrote: > > Hi, > Does anyone have any good empirical data suggesting what the maximum > number of PRI calls (incoming and outgoing) > without hardware echo cancellation can be handled on a single box is? > I have a TE410P T1 (1st gen) card and I'm seeing interesting errors of > D-Channels going down and then coming back up (See below). > > I've looked at the number of simultaneous calls at each of these > points, and each time the span seems to > have around 21-23 calls, and the total number of calls ranges between 47 and > 53. > I'm trying to figure out if this is a load issue or an issue on the > provider side, though my provider says they > do not see any errors on any of the T1s. > Could this be some sort of hardware interrupt problem? If so, how can I > check? > > The specs of the machine are, Dual Xeon 2.80Ghz (both single core but w/HT) > 4GB memory. > Running asterisk 1.4.26.3 (32-bit) > with libpri-1.4.7 and zaptel-1.4.12.9 > > Thanks. > > -- James > > Please CC me on responses. > > > [Mar 22 09:45:00] VERBOSE[8887] logger.c: == Primary D-Channel on span 2 > down > [Mar 22 09:45:00] WARNING[8887] chan_dahdi.c: No D-channels available! > Using Primary channel 48 as D-channel anyway! > [Mar 22 09:45:00] VERBOSE[8887] logger.c: == Primary D-Channel on span 2 > up > [Mar 22 09:59:23] VERBOSE[8886] logger.c: == Primary D-Channel on span 1 > down > [Mar 22 09:59:23] WARNING[8886] chan_dahdi.c: No D-channels available! > Using Primary channel 24 as D-channel anyway! > [Mar 22 09:59:23] VERBOSE[8886] logger.c: == Primary D-Channel on span 1 > up > [Mar 22 09:59:23] VERBOSE[8886] logger.c: == Primary D-Channel on span 1 > down > [Mar 22 09:59:23] WARNING[8886] chan_dahdi.c: No D-channels available! > Using Primary channel 24 as D-channel anyway! > [Mar 22 09:59:23] VERBOSE[8886] logger.c: == Primary D-Channel on span 1 > up > [Mar 22 10:36:11] VERBOSE[8886] logger.c: == Primary D-Channel on span 1 > down > [Mar 22 10:36:11] WARNING[8886] chan_dahdi.c: No D-channels available! > Using Primary channel 24 as D-channel anyway! > [Mar 22 10:36:11] VERBOSE[8886] logger.c: == Primary D-Channel on span 1 > up > [Mar 22 10:36:11] VERBOSE[8886] logger.c: == Primary D-Channel on span 1 > down > [Mar 22 10:36:11] WARNING[8886] chan_dahdi.c: No D-channels available! > Using Primary channel 24 as D-channel anyway! > [Mar 22 10:36:11] VERBOSE[8886] logger.c: == Primary D-Channel on span 1 > up > [Mar 22 10:44:36] NOTICE[] chan_dahdi.c: PRI got event: HDLC Bad > FCS (8) on Primary D-channel of span 3 > [Mar 22 10:45:44] NOTICE[8886] chan_dahdi.c: PRI got event: HDLC Bad > FCS (8) on Primary D-channel of span 1 > [Mar 22 10:59:33] NOTICE[8887] chan_dahdi.c: PRI got event: HDLC Abort > (6) on Primary D-channel of span 2 > [Mar 22 11:30:53] VERBOSE[8886] logger.c: == Primary D-Channel on span 1 > down > [Mar 22 11:30:53] WARNING[8886] chan_dahdi.c: No D-channels available! > Using Primary channel 24 as D-channel anyway! > [Mar 22 11:30:53] VERBOSE[8886] logger.c: == Primary D-Channel on span 1 > up > [Mar 22 15:34:28] VERBOSE[8887] logger.c:
Re: [asterisk-users] Maximum number of PRI calls on 1 asterisk box (no HW echo)
In theory asterisk can handle 8 spans, which means 192 concurrent calls if all spans are T1. In reality after 6 spans it starts giving problems. I recently shipped a server to a client in Taiwan, and before shipment it was put under a load of 100 plus concurrent calls (zap channels) for two whole weeks without any break. It had 6 spans, Rhino T1, FXO and FXS cards all with on board EC. The server's processors combined were hardly 7% in use and no call quality issues. For the same client another server with 9 spans were causing asterisk to freeze once there are more than 96 calls, i.e. it going to 1st span of the second T1 card, which made it 9th span considering FXO and FXS spans, but server load remained low as usual. So I think it is not your T1 card but some software/driver issue. Did you upgrade anything recently on this server? -- Zeeshan A Zakaria On 2010-03-24 8:49 PM, "James Lamanna" wrote: Hi, Does anyone have any good empirical data suggesting what the maximum number of PRI calls (incoming and outgoing) without hardware echo cancellation can be handled on a single box is? I have a TE410P T1 (1st gen) card and I'm seeing interesting errors of D-Channels going down and then coming back up (See below). I've looked at the number of simultaneous calls at each of these points, and each time the span seems to have around 21-23 calls, and the total number of calls ranges between 47 and 53. I'm trying to figure out if this is a load issue or an issue on the provider side, though my provider says they do not see any errors on any of the T1s. Could this be some sort of hardware interrupt problem? If so, how can I check? The specs of the machine are, Dual Xeon 2.80Ghz (both single core but w/HT) 4GB memory. Running asterisk 1.4.26.3 (32-bit) with libpri-1.4.7 and zaptel-1.4.12.9 Thanks. -- James Please CC me on responses. [Mar 22 09:45:00] VERBOSE[8887] logger.c: == Primary D-Channel on span 2 down [Mar 22 09:45:00] WARNING[8887] chan_dahdi.c: No D-channels available! Using Primary channel 48 as D-channel anyway! [Mar 22 09:45:00] VERBOSE[8887] logger.c: == Primary D-Channel on span 2 up [Mar 22 09:59:23] VERBOSE[8886] logger.c: == Primary D-Channel on span 1 down [Mar 22 09:59:23] WARNING[8886] chan_dahdi.c: No D-channels available! Using Primary channel 24 as D-channel anyway! [Mar 22 09:59:23] VERBOSE[8886] logger.c: == Primary D-Channel on span 1 up [Mar 22 09:59:23] VERBOSE[8886] logger.c: == Primary D-Channel on span 1 down [Mar 22 09:59:23] WARNING[8886] chan_dahdi.c: No D-channels available! Using Primary channel 24 as D-channel anyway! [Mar 22 09:59:23] VERBOSE[8886] logger.c: == Primary D-Channel on span 1 up [Mar 22 10:36:11] VERBOSE[8886] logger.c: == Primary D-Channel on span 1 down [Mar 22 10:36:11] WARNING[8886] chan_dahdi.c: No D-channels available! Using Primary channel 24 as D-channel anyway! [Mar 22 10:36:11] VERBOSE[8886] logger.c: == Primary D-Channel on span 1 up [Mar 22 10:36:11] VERBOSE[8886] logger.c: == Primary D-Channel on span 1 down [Mar 22 10:36:11] WARNING[8886] chan_dahdi.c: No D-channels available! Using Primary channel 24 as D-channel anyway! [Mar 22 10:36:11] VERBOSE[8886] logger.c: == Primary D-Channel on span 1 up [Mar 22 10:44:36] NOTICE[] chan_dahdi.c: PRI got event: HDLC Bad FCS (8) on Primary D-channel of span 3 [Mar 22 10:45:44] NOTICE[8886] chan_dahdi.c: PRI got event: HDLC Bad FCS (8) on Primary D-channel of span 1 [Mar 22 10:59:33] NOTICE[8887] chan_dahdi.c: PRI got event: HDLC Abort (6) on Primary D-channel of span 2 [Mar 22 11:30:53] VERBOSE[8886] logger.c: == Primary D-Channel on span 1 down [Mar 22 11:30:53] WARNING[8886] chan_dahdi.c: No D-channels available! Using Primary channel 24 as D-channel anyway! [Mar 22 11:30:53] VERBOSE[8886] logger.c: == Primary D-Channel on span 1 up [Mar 22 15:34:28] VERBOSE[8887] logger.c: == Primary D-Channel on span 2 down [Mar 22 15:34:28] WARNING[8887] chan_dahdi.c: No D-channels available! Using Primary channel 48 as D-channel anyway! [Mar 22 15:34:28] VERBOSE[8887] logger.c: == Primary D-Channel on span 2 up -- _ -- Bandwidth and Colocation Provided by http://www.api-digital.com -- New to Asterisk? Join us for a live introductory webinar every Thurs: http://www.asterisk.org/hello asterisk-users mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users -- _ -- Bandwidth and Colocation Provided by http://www.api-digital.com -- New to Asterisk? Join us for a live introductory webinar every Thurs: http://www.asterisk.org/hello asterisk-users mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users
[asterisk-users] Maximum number of PRI calls on 1 asterisk box (no HW echo)
Hi, Does anyone have any good empirical data suggesting what the maximum number of PRI calls (incoming and outgoing) without hardware echo cancellation can be handled on a single box is? I have a TE410P T1 (1st gen) card and I'm seeing interesting errors of D-Channels going down and then coming back up (See below). I've looked at the number of simultaneous calls at each of these points, and each time the span seems to have around 21-23 calls, and the total number of calls ranges between 47 and 53. I'm trying to figure out if this is a load issue or an issue on the provider side, though my provider says they do not see any errors on any of the T1s. Could this be some sort of hardware interrupt problem? If so, how can I check? The specs of the machine are, Dual Xeon 2.80Ghz (both single core but w/HT) 4GB memory. Running asterisk 1.4.26.3 (32-bit) with libpri-1.4.7 and zaptel-1.4.12.9 Thanks. -- James Please CC me on responses. [Mar 22 09:45:00] VERBOSE[8887] logger.c: == Primary D-Channel on span 2 down [Mar 22 09:45:00] WARNING[8887] chan_dahdi.c: No D-channels available! Using Primary channel 48 as D-channel anyway! [Mar 22 09:45:00] VERBOSE[8887] logger.c: == Primary D-Channel on span 2 up [Mar 22 09:59:23] VERBOSE[8886] logger.c: == Primary D-Channel on span 1 down [Mar 22 09:59:23] WARNING[8886] chan_dahdi.c: No D-channels available! Using Primary channel 24 as D-channel anyway! [Mar 22 09:59:23] VERBOSE[8886] logger.c: == Primary D-Channel on span 1 up [Mar 22 09:59:23] VERBOSE[8886] logger.c: == Primary D-Channel on span 1 down [Mar 22 09:59:23] WARNING[8886] chan_dahdi.c: No D-channels available! Using Primary channel 24 as D-channel anyway! [Mar 22 09:59:23] VERBOSE[8886] logger.c: == Primary D-Channel on span 1 up [Mar 22 10:36:11] VERBOSE[8886] logger.c: == Primary D-Channel on span 1 down [Mar 22 10:36:11] WARNING[8886] chan_dahdi.c: No D-channels available! Using Primary channel 24 as D-channel anyway! [Mar 22 10:36:11] VERBOSE[8886] logger.c: == Primary D-Channel on span 1 up [Mar 22 10:36:11] VERBOSE[8886] logger.c: == Primary D-Channel on span 1 down [Mar 22 10:36:11] WARNING[8886] chan_dahdi.c: No D-channels available! Using Primary channel 24 as D-channel anyway! [Mar 22 10:36:11] VERBOSE[8886] logger.c: == Primary D-Channel on span 1 up [Mar 22 10:44:36] NOTICE[] chan_dahdi.c: PRI got event: HDLC Bad FCS (8) on Primary D-channel of span 3 [Mar 22 10:45:44] NOTICE[8886] chan_dahdi.c: PRI got event: HDLC Bad FCS (8) on Primary D-channel of span 1 [Mar 22 10:59:33] NOTICE[8887] chan_dahdi.c: PRI got event: HDLC Abort (6) on Primary D-channel of span 2 [Mar 22 11:30:53] VERBOSE[8886] logger.c: == Primary D-Channel on span 1 down [Mar 22 11:30:53] WARNING[8886] chan_dahdi.c: No D-channels available! Using Primary channel 24 as D-channel anyway! [Mar 22 11:30:53] VERBOSE[8886] logger.c: == Primary D-Channel on span 1 up [Mar 22 15:34:28] VERBOSE[8887] logger.c: == Primary D-Channel on span 2 down [Mar 22 15:34:28] WARNING[8887] chan_dahdi.c: No D-channels available! Using Primary channel 48 as D-channel anyway! [Mar 22 15:34:28] VERBOSE[8887] logger.c: == Primary D-Channel on span 2 up -- _ -- Bandwidth and Colocation Provided by http://www.api-digital.com -- New to Asterisk? Join us for a live introductory webinar every Thurs: http://www.asterisk.org/hello asterisk-users mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users
